From 8cc826cbbcfda86432c5dfcae0380dfa54dbe8f0 Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 09:15:17 +0000 Subject: [PATCH 01/14] Add the FORMALIZED verification rung: a Lean-certificate lane above VERIFIED docs/FORMALIZE.md defines when a skeptic-confirmed proof step earns a machine-checked certificate, the statement-review step that carries the informal-to-formal bridge risk, and three pilot candidates from the current ledger. FORMALIZED enters the schema enum, the AGENTS.md vocabulary, the site badge/gloss tables, and the attempt template; problems/*/formal/ is registered tier 1. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- AGENTS.md | 1 + docs/FORMALIZE.md | 94 ++++++++++++++++++++++++++++++++++++++ docs/attempt-template.md | 3 +- docs/prior-art.schema.json | 3 +- scripts/build_site.py | 5 ++ tiers.json | 2 + 6 files changed, 106 insertions(+), 2 deletions(-) create mode 100644 docs/FORMALIZE.md diff --git a/AGENTS.md b/AGENTS.md index ee37551..139e4e1 100644 --- a/AGENTS.md +++ b/AGENTS.md @@ -99,6 +99,7 @@ Use these exact words; they are what the indexes are filtered on. | Term | Means | |---|---| | `VERIFIED` | Independently re-derived or re-computed. State the range or scope. | +| `FORMALIZED` | A verified proof step additionally carrying a machine-checked certificate (Lean 4). Covers exactly the formal statement, which is stated in the record. | | `EVIDENCE` | Computational support, not proof. State the range. | | `LIVE` | A route that has survived every adversary tried, with no bound claimed yet. | | `SPECULATION` | Plausible, unproven, and load-bearing. Must be labelled inline. | diff --git a/docs/FORMALIZE.md b/docs/FORMALIZE.md new file mode 100644 index 0000000..6241367 --- /dev/null +++ b/docs/FORMALIZE.md @@ -0,0 +1,94 @@ +# Formal certificates: the rung above the skeptic + +Adopted 2026-08-04, after OpenAI's Astra announcement (2026-08-02): ten +results on long-open problems, each shipped with a model-written reasoning +walkthrough *and* a Lean 4 certificate with a sorry-count of zero. Whatever +the eventual peer-review verdict on the individual results, the shape of the +release is the lesson: what made the claims credible — seven months after the +same company's "GPT-5 solved ten Erdős problems" collapsed into literature +rediscovery — was a kernel-checked artifact whose verdict does not depend on +trusting the model that produced it. + +That is the same role our skeptic pass plays, with a stronger guarantee for +one class of claim. This doc defines when and how a result here earns the +`FORMALIZED` status. + +## Where it sits + +| Rung | What it establishes | +|---|---| +| claim | nothing (an agent said so) | +| skeptic-confirmed (`VERIFIED`) | an independent adversary re-derived / re-implemented it and failed to kill it | +| `FORMALIZED` | a proof checker's kernel accepts a formal statement of it | + +Formalization is **optional and expensive**. It does not replace the skeptic +pass — it comes *after* it, and only for claims worth the cost. + +## What qualifies + +- **Proof-shaped claims only.** Lemmas, identities, case analyses, + monotonicity facts. Computational censuses do not gain from Lean: their + verification standard is already an independent re-implementation in exact + arithmetic (see the 2026-07-31 verification-standard note in `STATUS.md`), + and formalizing "the search was exhaustive" is a different, much larger job + than formalizing a lemma. +- **Skeptic-confirmed first.** Formalizing an unreviewed claim wastes the + expensive step on statements that a cheap adversary pass might kill. +- **Load-bearing and reused.** Prefer steps that many later records lean on; + a formalized wrong-turn is effort spent making a dead end rigorous. + +## The bridge caveat — read this before celebrating + +A Lean certificate proves exactly the **formal statement**, not the claim in +the attempt record. The informal→formal translation is the new risk surface, +and it is precisely the "hand-derived bridge" failure mode this lab already +recorded for formal-parameter specialization (STATUS insight, 2026-07-31): +the formal check cannot see a wrong bridge. Astra's coverage has the same +gap — a compiling proof of a mis-stated theorem is worth nothing. + +So a `FORMALIZED` pass has three parts, and the first is the point: + +1. **Statement review.** An independent agent (not the formalizer) compares + the formal statement, hypothesis by hypothesis, against the claim as + written in the attempt record, and says so explicitly in the review + record. Any narrowing (a stronger hypothesis, a special case) is listed + the way `range` is listed for a search. +2. **Clean build.** Zero `sorry`, no axioms beyond the standard library / + mathlib, and the full source in the repo. +3. **Recorded toolchain.** Lean toolchain version, mathlib pin, and the + `lake build` output captured in the record. CI does not build Lean (cost); + the reproducibility contract is that the skeptic's own independent + `lake build` — on a different machine or container — is part of the pass. + +## Mechanics + +- Files live in `problems//formal/` (tier 1; a blind agent never + sees them). +- Record the work as a **review-shape attempt** (`verifies: ` in the + index) with `status: FORMALIZED`. The scope line states the formal theorem + proved, which may be narrower than the record's claim — that difference is + data, not a footnote. +- `FORMALIZED` is in the schema's attempt-status enum and the `AGENTS.md` + vocabulary. Route-level status stays `VERIFIED`/`LIVE`/etc.; formalization + upgrades a *step*, not a conjecture. +- Lean is an exception to the stdlib-only-Python rule (that rule governs the + Python tooling). Nothing outside `formal/` may depend on it. + +## Pilot candidates (2026-08-04) + +Ordered by (value ÷ effort), from the current ledger: + +1. **Billiards Laurent-identity block** (005/008): I1–I4 and the glide facts + are already "one exact polynomial check" after adjoining e^{iaα}, e^{ibβ} + as formal variables — the closest thing in the ledger to a proof that is + already formal. Formalizing the polynomial identity is the easy half; the + *bridge* (geometry = closed form) is exactly what statement review is for. +2. **The c·cot(cs) monotonicity lemma** (005/008): elementary, one-variable, + mathlib has the pieces; it retires Lemmas C and D. +3. **Union-closed P6′** (008/012): quantitative IFT for the symmetric + Sinkhorn potential. Hardest of the three; attempt only after 1–2 have + established the workflow. + +A pilot that fails is recorded like any other attempt: what resisted +formalization, and whether the obstruction was the mathematics or the +tooling, is the deliverable. diff --git a/docs/attempt-template.md b/docs/attempt-template.md index 47afd40..fcdcd3c 100644 --- a/docs/attempt-template.md +++ b/docs/attempt-template.md @@ -40,7 +40,8 @@ Enough that a reader can follow and check it. Include: ## Outcome -Lead with a status term: `VERIFIED` / `EVIDENCE` / `LIVE` / `REFUTED` / `MAP`. +Lead with a status term: `VERIFIED` / `EVIDENCE` / `LIVE` / `REFUTED` / `MAP` +(reviews that produce a machine-checked certificate use `FORMALIZED`). State the **range or scope** — `VERIFIED` describes what was checked, never the conjecture itself. diff --git a/docs/prior-art.schema.json b/docs/prior-art.schema.json index 4e1feca..6735a1c 100644 --- a/docs/prior-art.schema.json +++ b/docs/prior-art.schema.json @@ -46,7 +46,8 @@ "status": { "enum": [ "VERIFIED", "VERIFIED_WITH_CORRECTIONS", "EVIDENCE", "LIVE", - "REFUTED", "REFUTED_PRIOR_OBSERVATION", "MAP", "SPECULATION" + "REFUTED", "REFUTED_PRIOR_OBSERVATION", "MAP", "SPECULATION", + "FORMALIZED" ] }, "mechanism": { diff --git a/scripts/build_site.py b/scripts/build_site.py index 6445a6f..5283ded 100644 --- a/scripts/build_site.py +++ b/scripts/build_site.py @@ -49,6 +49,7 @@ def issue_url(template: str, **prefill: str) -> str: STATUS = { "VERIFIED": ("verified", "Verified"), "VERIFIED_WITH_CORRECTIONS": ("corrected", "Verified, with corrections"), + "FORMALIZED": ("verified", "Formalized"), "EVIDENCE": ("evidence", "Evidence"), "LIVE": ("live", "Live"), "REFUTED": ("refuted", "Refuted"), @@ -66,6 +67,10 @@ def issue_url(template: str, **prefill: str) -> str: "did it also found errors in the record it was checking. The original is " "kept exactly as written; the corrections are listed against it. Read the " "corrections before building on the claim.", + "FORMALIZED": "A verified proof step additionally carrying a machine-checked " + "certificate (Lean 4), so its verdict does not depend on trusting whoever " + "produced it. Covers exactly the formal statement, which the record states " + "and which may be narrower than the informal claim.", "EVIDENCE": "Computational support, not proof. A search to 106 is " "evidence about 106.", "LIVE": "A route that has survived every adversary tried, with no bound yet " diff --git a/tiers.json b/tiers.json index c9b47a1..69d45a0 100644 --- a/tiers.json +++ b/tiers.json @@ -20,11 +20,13 @@ "mechanisms.json", "docs/IDEATE.md", "docs/RIPPLE.md", + "docs/FORMALIZE.md", "docs/PLAN-physics-problems.md", ".claude/**/*", "problems/*/PRIOR-ART.md", "problems/*/prior-art.json", "problems/*/attempts/**/*", + "problems/*/formal/**/*", "problems/*/explore/**/*", "problems/*/data/**/*", "site/**/*", From 8513c6a3419f370792556bab553905e1534be8bf Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 09:15:27 +0000 Subject: [PATCH 02/14] Wire the literature-novelty gate and formalization rung into the procedure The skeptic pass now checks claimed-new results against the literature (rediscovery is recorded as rediscovery, with the citation), and CONTRIBUTING.md carries it as bar item 7. CYCLE.md step 4 points at the FORMALIZED escalation for load-bearing proof-shaped survivors. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- CONTRIBUTING.md | 7 ++++++- docs/CYCLE.md | 14 +++++++++++++- 2 files changed, 19 insertions(+), 2 deletions(-) diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index 217fc58..4f11e1d 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -28,9 +28,14 @@ an AI-generated mathematics repo that skips them is worse than nothing. 6. **Transcription is marked.** If a source detail came from a machine transcription of a paper rather than the paper, mark it and say so, as attempt 003 does with `[L]`. +7. **Novelty is checked.** A result recorded as *new* has been searched for + in the literature first. Rediscovering a published result is a fine + outcome — record it as a rediscovery, with the citation. Claiming a known + result as new is the failure mode. Use the status vocabulary in `AGENTS.md` exactly: `VERIFIED`, `EVIDENCE`, -`LIVE`, `SPECULATION`, `REFUTED`, `MAP`. +`LIVE`, `SPECULATION`, `REFUTED`, `MAP` — plus `FORMALIZED` for a verified +proof step that additionally carries a machine-checked certificate. ## Submitting an attempt diff --git a/docs/CYCLE.md b/docs/CYCLE.md index 170f20d..8a5744e 100644 --- a/docs/CYCLE.md +++ b/docs/CYCLE.md @@ -70,7 +70,14 @@ refute**. Not "check this" — *try to kill it*. The skeptic must: of one program agreeing tells you nothing about whether the program is right; - check that headline constants come from the actual extremum, not from where a search grid happened to stop; -- verify that cited sources say what they are claimed to say. +- verify that cited sources say what they are claimed to say; +- **check claimed novelty against the literature.** Search for the statement + and its neighborhood before any record calls a finding *new*. Rediscovering + a published result is a fine outcome — blind-mode rediscovery is calibration + data this lab wants — but it is recorded as a rediscovery, with the + citation. The failure mode is the October 2025 "GPT-5 solved ten Erdős + problems" episode: real derivations, already in the literature, announced + as new. If the skeptic finds something, that finding is recorded, and the original record is **left as written**. See `problems/union-closed/attempts/004-*` for @@ -78,6 +85,11 @@ what a good skeptic pass looks like: it confirmed the main interface, refuted a mini-theorem, corrected a headline constant, and flagged a data file that could not be reproduced. +For a proof-shaped claim that survives the skeptic and is load-bearing, there +is one rung above this: a machine-checked certificate (`docs/FORMALIZE.md`, +status `FORMALIZED`). Optional and expensive — reserve it for steps many later +records will lean on, and read the bridge caveat there first. + ### 5. Write it up Every line pursued gets a record, whether it worked or not: From 442cfdd9d5cd8ca34856f3ce5f3edeb77489fe2a Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 09:15:27 +0000 Subject: [PATCH 03/14] Guidance and ledger: direction from the 2026-08-02 Astra announcement New current-guidance section (2026-08-04): ripple-scan the ten Astra results before other queue work, novelty gate and formalization lane adopted, portfolio note on preferring decade-open specialist-tractable targets with checkable deliverables. STATUS.md gets the dated insights entry, queue item 15 (ripple scan) and 16 (billiards formalization pilot), and a TL;DR process note. No mathematical state changed. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- GUIDANCE.md | 39 ++++++++++++++++++++++++++++++++++++++- STATUS.md | 25 ++++++++++++++++++++++++- 2 files changed, 62 insertions(+), 2 deletions(-) diff --git a/GUIDANCE.md b/GUIDANCE.md index dcc87f2..b39c0c5 100644 --- a/GUIDANCE.md +++ b/GUIDANCE.md @@ -4,7 +4,44 @@ Human-set direction for the lab. `docs/CYCLE.md` has agents read this at the start of every cycle, and **it overrides the queue** where the two disagree — so it has to describe how things actually are, not how they once were. -## Current guidance (2026-07-26) +## Current guidance (2026-08-04) + +Everything in the 2026-07-26 section below still stands — operating mode, +public-repo posture, and the calibration rules. This section adds direction +drawn from OpenAI's Astra announcement (2026-08-02): ten results on +decade-plus-open problems in mathematics and TCS, each published with a +model-written reasoning walkthrough and a Lean 4 certificate at zero sorries, +for a reported ~$2,000 of tokens. None are peer-reviewed yet; treat the +individual results as reported, not settled. The *shape* of the release is +what matters here, and three things follow for this lab: + +1. **Run the ripple scan first.** Before the next cycle pulls anything else + from the queue, run `docs/RIPPLE.md` on the Astra results (queue item 15). + The Ehrhart-volume and extremal-graph/Ramsey entries are the plausible + bites on mahler-4d and erdos-gyarfas; the scan decides, not this note. +2. **The novelty gate is now part of the bar.** The October 2025 "GPT-5 + solved ten Erdős problems" collapse was a labeling failure, not a + derivation failure: correct arguments, already in the literature, claimed + as new. Our blind mode *wants* rediscoveries — but recorded as + rediscoveries. The skeptic pass now includes a literature check on any + claimed-new result (`docs/CYCLE.md` step 4, `CONTRIBUTING.md` bar item 7). +3. **Formalization is the new top verification rung.** For skeptic-confirmed, + load-bearing, proof-shaped steps, a kernel-checked Lean certificate is the + one thing stronger than a second independent derivation — it is what made + Astra's claims land after the October episode. `docs/FORMALIZE.md` + defines the lane and the `FORMALIZED` status; queue item 16 is the pilot. + Expensive; use deliberately. + +**Portfolio note.** Astra's target list — decade-open but +specialist-tractable problems whose deliverables are checkable objects +(constructions, counterexamples, quantitative bound improvements) — is a +validation of this lab's existing tilt toward the smaller open problems, not +a reason to chase the famous ones harder. When onboarding new problems, +prefer that shape: a problem whose answer, if found, is an object a verifier +can check (the Erdős catalog is a good source). Famous long shots keep their +deliberate minority share. + +## Previous guidance (2026-07-26) **Operating mode.** No continuous loop is running. Work happens on demand: a cycle is run deliberately, by this lab or by an outside contributor's agent, diff --git a/STATUS.md b/STATUS.md index 4083355..1ffe79e 100644 --- a/STATUS.md +++ b/STATUS.md @@ -3,7 +3,7 @@ The live ledger: where every problem stands, what is queued next, and what has already been ruled out. Read this before starting work. -## TL;DR (updated 2026-07-31) +## TL;DR (updated 2026-08-04) **Open for contributions. No automated loop is currently running** — the hourly cycle that produced these records was stopped once its in-flight work closed @@ -11,6 +11,13 @@ out, and no cron trigger is active. Nothing is half-finished: every line is written up, and the queue below is a list of starting points rather than abandoned work. +**Process update (2026-08-04), prompted by OpenAI's Astra announcement:** +a literature-novelty check is now part of the skeptic pass, a formalization +lane (`docs/FORMALIZE.md`, status `FORMALIZED`) sits above `VERIFIED` for +load-bearing proof steps, and `GUIDANCE.md` directs the next cycle to start +with a ripple scan of the Astra results (queue items 15–16). No mathematical +state changed. + To pick something up, take an item from the attempt queue and follow `docs/CYCLE.md` — either by hand, one line at a time, or by restarting an hourly routine with that file as the prompt. Everything needed is in this @@ -80,6 +87,8 @@ attempts (queue; run blind). 12. [billiards-triangles] **The coverage conjecture, and the sampler blind spot** (from 006/007, 2026-07-31; absorbs the old pinch-gap item — its motivating gap [135.000°, 135.049°] is CLOSED, W(4,3) is certified alive inside it): 006 reduced "every obtuse angle has an alive W member" to an elementary Diophantine statement (unproven; float-checked at 157 + 25 arcs over 90.5°–165°, zero failures). Prove it, using the birth law as a labelled input where needed. Note the certificates so far are POINTWISE (007's C2): window-interval continuity on sub-arcs is float + SPECULATION law only, and per-triangle coverage of a whole arc is a different (open) question — the windows are x-slivers at the corners. Separately falsifiable (007 lead): every sampler in use accumulates only at the 90/j window edges, so an interior-pinch alive window would hide from ALL current designs — build one targeted interior-accumulation test before trusting any negative screen again. 13. [mahler-4d] Close the {0,±1}⁴ universe: k = 12–20 pairs (~30M orbits at k=12, improper fraction already 77% at k=9). Falsifiable: no proper mask with k ≥ 12 has P < 11. Needs the improper-detection shortcut or a streaming canonicalizer; see 001 lead 1. Cheap side quest, same pipeline: the {0,±1}³ census for the n=3 spectrum comparison (13 pairs, trivial) — does the non-Hanner gap grow or shrink with n? 14. [billiards-triangles] Coverage self-test: re-derive the acute and right-triangle cases as a scoped attempt record. Low value now that the harness self-test covers Fagnano and the orthic geometry and 001 mapped the obtuse side — take it only if something turns up that the certificate machinery cannot express. +15. [cross-problem] **Ripple scan the Astra results** (GUIDANCE 2026-08-04 pulls this to the front of the next cycle). Per `docs/RIPPLE.md`: characterize each of the ten 2026-08-02 OpenAI results (scope, mechanism tags, what the argument needs to apply), then scan the portfolio. Plausible bites to check first, without prejudging: the Ehrhart volume-conjecture proof against mahler-4d (both live in lattice-polytope volume land — does its machinery say anything about volume products, or vice versa?), and the extremal-graph-theory / multicolor-Ramsey entries against erdos-gyarfas (cycle-spectrum arguments share the field lens). Verify each result's actual statement from the published manuscripts/Lean certificates, not press coverage — these are unrefereed, so scope-check harder than usual, and treat any transfer hypothesis as SPECULATION. Expected outcome: recorded misses; a hit goes to the top of that problem's queue; a scan that changes the picture becomes a MAP attempt. +16. [billiards-triangles] **Formalization pilot** (`docs/FORMALIZE.md`): produce the lab's first Lean 4 certificate for the c·cot(cs) monotonicity lemma (elementary, retires Lemmas C and D), then attempt the Laurent-identity block I1–I4 (already one exact polynomial check — but its geometry bridge is exactly what the statement-review step exists for). Record as a review-shape attempt with `status: FORMALIZED`; toolchain, mathlib pin, and `lake build` output in the record; independent statement review by a non-formalizer agent is part of the pass. A failed pilot is a full deliverable if it pins whether the obstruction is mathematical or tooling. ## Verified results @@ -314,6 +323,20 @@ attempts (queue; run blind). ## Insights / cross-problem notes +- External event + process change (2026-08-04): OpenAI announced (2026-08-02) + ten results on decade-plus-open problems (non-sofic groups, Connes rigidity + counterexample, Ehrhart volume conjecture, three Erdős-catalog problems + incl. multicolor Ramsey #183, sphere-packing and coding bounds, arithmetic + circuits, quantum parallel repetition, closest-vector, extremal graphs), + each with a model reasoning walkthrough and a Lean 4 certificate at zero + sorries; unrefereed as of this note. Adopted here: (1) literature-novelty + check added to the skeptic pass — the instructive contrast is OpenAI's own + October 2025 episode, where correct derivations already in the literature + were announced as new; (2) `FORMALIZED` status + `docs/FORMALIZE.md` lane + for kernel-checked certificates above `VERIFIED`, with the statement-review + step carrying the bridge risk we already hit in formal-parameter + specialization; (3) ripple scan of the ten results queued as item 15, + formalization pilot as item 16. No mathematical state changed by this note. - Infrastructure (2026-07-27): cross-pollination layer added. `mechanisms.json` maps every approach tag to a field lens (`scripts/mechanisms.py` for gaps/matrix queries); `docs/IDEATE.md` = field-sweep ideation on one From 8a732c936c2c1fcebe014f4f5fe48c690233fb77 Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 09:36:47 +0000 Subject: [PATCH 04/14] billiards-triangles 009: Lean pilot - Lemma L1 machine-checked, statement review pending First certificate of the docs/FORMALIZE.md lane: 005's Lemma L1 (for fixed s in (0, pi/2], c |-> c*cot(cs) strictly decreasing on (0,1]) proved in Lean 4 / mathlib v4.32.2, zero sorry, axioms limited to propext/Classical.choice/Quot.sound, prebuilt mathlib via lake cache. Review-shape record with the hypothesis mapping flagged pending independent statement review; formal/.lake is git-ignored. New mechanism tag machine-checked-formalization. test_records.py now skips .lake/ vendored docs in the missing-file lint. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- mechanisms.json | 4 + .../attempts/009-lean-pilot-cot-lemma.md | 172 ++++++++++++++++++ .../billiards-triangles/formal/.gitignore | 5 + .../formal/BilliardsFormal.lean | 1 + .../formal/BilliardsFormal/CotMonotone.lean | 96 ++++++++++ .../billiards-triangles/formal/lake-build.log | 10 + .../formal/lake-manifest.json | 96 ++++++++++ .../billiards-triangles/formal/lakefile.toml | 10 + .../billiards-triangles/formal/lean-toolchain | 1 + problems/billiards-triangles/prior-art.json | 26 +++ tests/test_records.py | 4 +- 11 files changed, 424 insertions(+), 1 deletion(-) create mode 100644 problems/billiards-triangles/attempts/009-lean-pilot-cot-lemma.md create mode 100644 problems/billiards-triangles/formal/.gitignore create mode 100644 problems/billiards-triangles/formal/BilliardsFormal.lean create mode 100644 problems/billiards-triangles/formal/BilliardsFormal/CotMonotone.lean create mode 100644 problems/billiards-triangles/formal/lake-build.log create mode 100644 problems/billiards-triangles/formal/lake-manifest.json create mode 100644 problems/billiards-triangles/formal/lakefile.toml create mode 100644 problems/billiards-triangles/formal/lean-toolchain diff --git a/mechanisms.json b/mechanisms.json index 71ed16a..3cc92c2 100644 --- a/mechanisms.json +++ b/mechanisms.json @@ -152,6 +152,10 @@ "field": "extremal-combinatorics", "description": "Generalize a two-element union/combination argument to k-wise combinations and track how constants move." }, + "machine-checked-formalization": { + "field": "logic-methodology", + "description": "State a skeptic-confirmed proof step in a proof assistant (Lean 4 + mathlib) and have the kernel accept a sorry-free proof; the informal-to-formal statement bridge is reviewed separately per docs/FORMALIZE.md." + }, "mordell-qr-obstruction": { "field": "number-theory", "description": "Use quadratic-residue conditions to characterize which residue classes a covering-congruence argument can reach." diff --git a/problems/billiards-triangles/attempts/009-lean-pilot-cot-lemma.md b/problems/billiards-triangles/attempts/009-lean-pilot-cot-lemma.md new file mode 100644 index 0000000..2030265 --- /dev/null +++ b/problems/billiards-triangles/attempts/009-lean-pilot-cot-lemma.md @@ -0,0 +1,172 @@ +# 009 — Lean 4 formalization pilot: Lemma L1 of 005 machine-checked + +- **Problem:** billiards-triangles, `problems/billiards-triangles/PROBLEM.md` +- **Date:** 2026-08-04 +- **Mode:** informed (read `prior-art.json`, 005 §(ii)-B and front matter, + 008 §3, `docs/FORMALIZE.md`, `STATUS.md` queue item 16, `AGENTS.md`, + `CONTRIBUTING.md`) +- **Type:** formalization — review-shape attempt against + `005-complete-death-law-theorem.md` (verifies 005's Lemma L1 with a + kernel-checked Lean 4 certificate). This is the lab's first run of the + `docs/FORMALIZE.md` lane (queue item 16, first half; the Laurent-identity + block I1–I4 is NOT attempted here). +- **Tools:** elan 4.2.3; Lean 4 toolchain `leanprover/lean4:v4.32.2`; + mathlib pinned at tag `v4.32.2` + (rev `905b95818eb32af7874a58b427f50c1711a5e96c`); project at + `problems/billiards-triangles/formal/` (`lakefile.toml`, `lean-toolchain`, + `lake-manifest.json`, source `BilliardsFormal/CotMonotone.lean`). + Prebuilt mathlib via `lake exe cache get` (nothing of mathlib rebuilt from + source). Build is deterministic; the pilot's own module compiles in ~8 s; + build + axiom-audit output captured in `formal/lake-build.log`. +- **Sources:** none external beyond mathlib itself. Repo: 005 (the record + whose lemma is formalized), 008 (its skeptic review, §3 CONFIRMED), + `docs/FORMALIZE.md` (the lane definition, including the bridge caveat this + record's mapping section exists for). + +## Claims attacked + +1. **005 Lemma L1** (§(ii)-B, radians): "For fixed s ∈ (0, π/2], the map + c ↦ c·cot(cs) is strictly decreasing on (0, 1]." Already + skeptic-confirmed by hand in 008 §3; the attack here is the stronger + one the FORMALIZE lane defines: state it formally and make Lean's kernel + accept a proof. A false or mis-stated lemma would have refused to close. +2. The **informal→formal bridge** for that statement: that the formal + theorem proved is the record's sentence and not a narrowing (the failure + surface `docs/FORMALIZE.md` warns about). Attacked by a + hypothesis-by-hypothesis mapping, written out below — but per the lane, + the mapping must be re-checked by an independent reviewer, so it is + flagged **PENDING INDEPENDENT STATEMENT REVIEW**. + +Not attacked: Lemmas C and D themselves (L1's consumers), the I1–I4 +Laurent block, and everything else in 005. No claim of 005 is re-derived +informally here; 008 already did that. + +## Refutations found + +None. The lemma formalized exactly as stated in the record: no hypothesis +had to be strengthened, no case dropped, no strictness weakened. (One +non-finding worth recording: mathlib at v4.32.2 has `Real.cot` and +`Real.cot_eq_cos_div_sin` but no derivative API for `Real.cot`, so the +quotient-rule step is done inline — see "what was needed" below. That is a +tooling observation, not a gap in 005.) + +## Claims that survive + +### The formal statement proved (verbatim from `formal/BilliardsFormal/CotMonotone.lean`) + +```lean +theorem cot_lemma_L1 {s : ℝ} (hs0 : 0 < s) (hs : s ≤ π / 2) : + StrictAntiOn (fun c : ℝ => c * Real.cot (c * s)) (Set.Ioc 0 1) +``` + +proved with zero `sorry`, via the intermediate theorem (same file) + +```lean +theorem strictAntiOn_id_mul_cot : + StrictAntiOn (fun u : ℝ => u * Real.cot u) (Set.Ioc 0 (π / 2)) +``` + +Axiom audit (in `formal/lake-build.log`): both theorems depend on exactly +`[propext, Classical.choice, Quot.sound]` — the three standard Lean/mathlib +axioms, nothing else, no `sorryAx`. + +### Mapping to 005's informal statement — **PENDING INDEPENDENT STATEMENT REVIEW** + +005 §(ii)-B: "**Lemma L1.** For fixed s ∈ (0, π/2], the map c ↦ c·cot(cs) +is strictly decreasing on (0, 1]." Hypothesis by hypothesis: + +| 005's informal statement | formal counterpart | delta | +|---|---|---| +| "fixed s ∈ (0, π/2]" | `{s : ℝ} (hs0 : 0 < s) (hs : s ≤ π / 2)` | none: (0, π/2] written as the two inequalities | +| "the map c ↦ c·cot(cs)" | `fun c : ℝ => c * Real.cot (c * s)` | none; `Real.cot` is radian, matching 005's "radians throughout this subsection" | +| "strictly decreasing on (0, 1]" | `StrictAntiOn … (Set.Ioc 0 1)` | none: `StrictAntiOn f I` is mathlib's "∀ c₁ c₂ ∈ I, c₁ < c₂ → f c₂ < f c₁", and `Set.Ioc 0 1` is (0, 1] | +| cot well-defined on the range | no side condition needed | for c ∈ (0,1], cs ∈ (0, π/2] where sin > 0; `Real.cot` is total (junk value where sin = 0) but the proof never evaluates it there | + +Deliberate proof-shape delta (statement unchanged): the Lean proof first +proves the substituted form `strictAntiOn_id_mul_cot` — u·cot u strictly +decreasing on (0, π/2] — and derives L1 by u = c·s (rescaling by the +positive constant s). That is 005's own proof reparametrized: 005 +differentiates in c and gets d/dc[c·cot(cs)] = [sin(2cs) − 2cs]/(2sin²(cs)); +the Lean proof differentiates in u = cs and gets +(sin u·cos u − u)/sin²u = [sin(2u) − 2u]/(2 sin²u), the identical +expression. The inequality driving it is the same `sin x < x for x > 0` +(mathlib `Real.sin_lt`). The theorem exported as `cot_lemma_L1` is the +record's statement, not the substituted one. + +What L1's certificate does **not** cover (unchanged from 005/008's own +scoping): Lemmas C and D are informal consumers of L1 — their derivations +(log-derivative split, endpoint identity, factorization; Lemma D's +b·cot(bx) − cot x < 0) remain at `VERIFIED` (008 §3 hand re-derivation), +not `FORMALIZED`. Formalizing them is the natural next increment; see +leads. + +### Toolchain and reproduction + +- `formal/lean-toolchain`: `leanprover/lean4:v4.32.2` +- mathlib: tag `v4.32.2` = rev `905b95818eb32af7874a58b427f50c1711a5e96c` + (pinned in `formal/lakefile.toml`, locked in `formal/lake-manifest.json`) +- Commands, from `problems/billiards-triangles/formal/`: + `lake exe cache get && lake build` — result: + `Build completed successfully (8657 jobs)`, log in `formal/lake-build.log` + together with the `#print axioms` audit of both theorems. +- Per `docs/FORMALIZE.md`, CI does not build Lean; the reproducibility + contract is a skeptic's own independent `lake build` on another machine. + That, plus the statement review above, completes the FORMALIZED pass — + neither is done by this record. + +### Pilot verdict on the lane itself (queue item 16 asked for this) + +The obstruction-vs-tooling question has an answer: **no obstruction, either +mathematical or tooling, for a lemma of this class.** Toolchain install to +green build was well under the timebox: elan install ~1 min, mathlib clone + +`lake exe cache get` ~25 min (≈7 GB in `.lake/`, git-ignored), the proof +itself compiled on the first `lake build`. The one friction point: mathlib +v4.32.2 has no `HasDerivAt` lemma for `Real.cot`, so the derivative is +computed inline by the quotient rule on cos/sin +(`hasDerivAt_id_mul_cot` in the file) — three lines, not a blocker. The +useful mathlib surface: `strictAntiOn_of_deriv_neg`, +`Real.hasDerivAt_sin/cos`, `HasDerivAt.mul/div`, `Real.sin_lt`, +`Real.cot_eq_cos_div_sin`, `linear_combination` for the Pythagorean +rewrite. + +## Leads generated + +1. **Complete the FORMALIZED pass on this record.** An independent agent + (not this one): (a) re-runs `lake build` in a fresh container from the + committed source (definite outcome: builds green or does not), and + (b) performs the statement review of the mapping table above against + 005 §(ii)-B (definite outcome: signs off or files a delta). Until both, + the certificate is a formalizer-claimed one. +2. **Formalize Lemma D** (sin(bx)/sin x strictly decreasing on (0, π/(2b)], + b ≥ 2): it is a two-line consumer of L1 (`b·cot(bx) − cot x < 0` via L1 + at s = bx, c = 1/b vs c = 1) plus a log-derivative wrapper. Definite + outcome: compiles against `cot_lemma_L1` or exposes a domain subtlety in + the b·x substitution. Cheap; would make the first *chain* L1 → D formal. +3. **Queue item 16, second half — the Laurent-identity block I1–I4.** + 005 proves these by exact polynomial subtraction in + ℤ[x, x⁻¹, y, y⁻¹, S, T]; the Lean analogue is an identity of Laurent + polynomials, decidable by `decide`/`ring` over ℤ. The geometry bridge + (u = R₀·Rot_A(2aα)·Rot_B(−2bβ)) is exactly what the statement-review + step must scrutinize there — the formal check cannot see a wrong bridge + (STATUS insight 2026-07-31). Definite outcome per identity: `ring` + closes it or the formalization surfaces a mismatch with + `plaw_general.py`'s ring. +4. **Lemma C** is the hard one of the pair (endpoint identity + π/2 = (b+1)v₀, factorization with sign analysis of each factor); attempt + only after 2 establishes the substitution pattern. Definite outcome: + `H2 > 0` on (0, v₀) formal, or a pinned list of which of 005's five + inequality steps resists mathlib's API. + +## References + +- `problems/billiards-triangles/attempts/005-complete-death-law-theorem.md` + §(ii)-B (Lemma L1, statement and informal proof; the record verified). +- `problems/billiards-triangles/attempts/008-skeptic-review-of-005.md` §3 + (hand re-derivation of L1/C/D, 14,183-point adversarial numerics — the + skeptic-confirmation that `docs/FORMALIZE.md` requires before + formalizing). +- `docs/FORMALIZE.md` (the lane: qualification, bridge caveat, three-part + pass). +- mathlib4, tag v4.32.2, github.com/leanprover-community/mathlib4 + (rev `905b95818eb32af7874a58b427f50c1711a5e96c`). No papers consulted; + no machine transcriptions. diff --git a/problems/billiards-triangles/formal/.gitignore b/problems/billiards-triangles/formal/.gitignore new file mode 100644 index 0000000..21cc577 --- /dev/null +++ b/problems/billiards-triangles/formal/.gitignore @@ -0,0 +1,5 @@ +# Lake build artifacts and downloaded dependencies / mathlib cache — never commit. +.lake/ +# Editor/OS noise +.DS_Store +*.olean diff --git a/problems/billiards-triangles/formal/BilliardsFormal.lean b/problems/billiards-triangles/formal/BilliardsFormal.lean new file mode 100644 index 0000000..cf2f6f4 --- /dev/null +++ b/problems/billiards-triangles/formal/BilliardsFormal.lean @@ -0,0 +1 @@ +import BilliardsFormal.CotMonotone diff --git a/problems/billiards-triangles/formal/BilliardsFormal/CotMonotone.lean b/problems/billiards-triangles/formal/BilliardsFormal/CotMonotone.lean new file mode 100644 index 0000000..28e1e45 --- /dev/null +++ b/problems/billiards-triangles/formal/BilliardsFormal/CotMonotone.lean @@ -0,0 +1,96 @@ +/- +Lemma L1 of attempt record 005 (`problems/billiards-triangles/attempts/ +005-complete-death-law-theorem.md`, section (ii)-B), reviewed and confirmed in +008 (section 3): + + **Lemma L1.** For fixed s ∈ (0, π/2], the map c ↦ c·cot(c·s) is strictly + decreasing on (0, 1]. + +This is the elementary monotonicity lemma that retires Lemmas C and D of the +W(a,b) death-law theorem (005 uses it three times: the two brackets of +Lemma C's log-derivative, and Lemma D). + +Formal statement proved here: `cot_lemma_L1` below — + `StrictAntiOn (fun c : ℝ => c * Real.cot (c * s)) (Set.Ioc 0 1)` +for every real `s` with `0 < s` and `s ≤ π / 2`. + +The proof follows 005's own two-line argument: the derivative of +`u ↦ u · cot u` is `(sin u · cos u − u) / sin² u < 0` (since `sin x < x` for +`x > 0`), giving strict decrease of `u · cot u` on `(0, π/2]` +(`strictAntiOn_id_mul_cot`); Lemma L1 follows by the substitution `u = c·s`, +which rescales the value by the positive constant `1/s`. + +Zero `sorry`; no axioms beyond mathlib. Toolchain pinned in `lean-toolchain` +(leanprover/lean4:v4.32.2), mathlib pinned in `lakefile.toml` / +`lake-manifest.json` (tag v4.32.2). +-/ +import Mathlib + +open Real Set + +namespace BilliardsFormal + +/-- The derivative of `u ↦ u * cos u / sin u` at a point where `sin x ≠ 0`. -/ +lemma hasDerivAt_id_mul_cot {x : ℝ} (hsin : Real.sin x ≠ 0) : + HasDerivAt (fun u : ℝ => u * Real.cos u / Real.sin u) + (((1 * Real.cos x + x * -Real.sin x) * Real.sin x + - x * Real.cos x * Real.cos x) / Real.sin x ^ 2) x := by + exact ((hasDerivAt_id' (x := x)).mul (Real.hasDerivAt_cos x)).div + (Real.hasDerivAt_sin x) hsin + +/-- Core monotonicity: `u ↦ u * cot u` is strictly decreasing on `(0, π/2]`. + (005's Lemma L1 after the substitution `u = c·s`.) -/ +theorem strictAntiOn_id_mul_cot : + StrictAntiOn (fun u : ℝ => u * Real.cot u) (Set.Ioc 0 (π / 2)) := by + have hfun : (fun u : ℝ => u * Real.cot u) + = fun u : ℝ => u * Real.cos u / Real.sin u := by + funext u + rw [Real.cot_eq_cos_div_sin, mul_div_assoc] + rw [hfun] + have hpi2 : (π / 2 : ℝ) < π := half_lt_self Real.pi_pos + have hsinpos : ∀ x ∈ Set.Ioc (0 : ℝ) (π / 2), 0 < Real.sin x := fun x hx => + Real.sin_pos_of_pos_of_lt_pi hx.1 (lt_of_le_of_lt hx.2 hpi2) + apply strictAntiOn_of_deriv_neg (convex_Ioc 0 (π / 2)) + · -- continuity on (0, π/2] + exact continuousOn_id.mul (Real.continuous_cos.continuousOn) |>.div + Real.continuous_sin.continuousOn (fun x hx => (hsinpos x hx).ne') + · -- negative derivative on the interior (0, π/2) + intro x hx + rw [interior_Ioc] at hx + have hx0 : 0 < x := hx.1 + have hsin : 0 < Real.sin x := hsinpos x ⟨hx.1, hx.2.le⟩ + rw [(hasDerivAt_id_mul_cot hsin.ne').deriv] + have heq : ((1 * Real.cos x + x * -Real.sin x) * Real.sin x + - x * Real.cos x * Real.cos x) / Real.sin x ^ 2 + = (Real.sin x * Real.cos x - x) / Real.sin x ^ 2 := by + linear_combination (-x / Real.sin x ^ 2) * Real.sin_sq_add_cos_sq x + rw [heq] + apply div_neg_of_neg_of_pos + · -- sin x · cos x < x, since sin x · cos x ≤ sin x < x + have h1 : Real.sin x * Real.cos x ≤ Real.sin x := + mul_le_of_le_one_right hsin.le (Real.cos_le_one x) + have h2 : Real.sin x < x := Real.sin_lt hx0 + linarith + · exact pow_pos hsin 2 + +/-- **Lemma L1** (record 005, section (ii)-B), stated exactly as written there: + for fixed `s ∈ (0, π/2]`, the map `c ↦ c·cot(c·s)` is strictly decreasing + on `(0, 1]`. `StrictAntiOn f I` is mathlib's "f is strictly decreasing on + I": for all `c₁, c₂ ∈ I` with `c₁ < c₂`, `f c₂ < f c₁`. -/ +theorem cot_lemma_L1 {s : ℝ} (hs0 : 0 < s) (hs : s ≤ π / 2) : + StrictAntiOn (fun c : ℝ => c * Real.cot (c * s)) (Set.Ioc 0 1) := by + intro c₁ hc₁ c₂ hc₂ hlt + have mem : ∀ c : ℝ, c ∈ Set.Ioc (0 : ℝ) 1 → c * s ∈ Set.Ioc (0 : ℝ) (π / 2) := + fun c hc => ⟨mul_pos hc.1 hs0, + le_trans (mul_le_of_le_one_left hs0.le hc.2) hs⟩ + have h := strictAntiOn_id_mul_cot (mem c₁ hc₁) (mem c₂ hc₂) + (mul_lt_mul_of_pos_right hlt hs0) + -- h : (c₂·s)·cot(c₂·s) < (c₁·s)·cot(c₁·s); divide by s > 0. + have h' : s * (c₂ * Real.cot (c₂ * s)) < s * (c₁ * Real.cot (c₁ * s)) := by + calc s * (c₂ * Real.cot (c₂ * s)) + = (c₂ * s) * Real.cot (c₂ * s) := by ring + _ < (c₁ * s) * Real.cot (c₁ * s) := h + _ = s * (c₁ * Real.cot (c₁ * s)) := by ring + exact lt_of_mul_lt_mul_left h' hs0.le + +end BilliardsFormal diff --git a/problems/billiards-triangles/formal/lake-build.log b/problems/billiards-triangles/formal/lake-build.log new file mode 100644 index 0000000..f8d1e11 --- /dev/null +++ b/problems/billiards-triangles/formal/lake-build.log @@ -0,0 +1,10 @@ +$ lake build # 2026-08-04T09:31:47Z +$ lean --version: Lean (version 4.32.2, x86_64-unknown-linux-gnu, commit f3b06c705e6c85f5314019d5d3baab0fec5b580c, Release) +$ mathlib rev: 905b95818eb32af7874a58b427f50c1711a5e96c (tag v4.32.2) +✔ [8655/8657] Built BilliardsFormal.CotMonotone (7.5s) +✔ [8656/8657] Built BilliardsFormal (5.7s) +Build completed successfully (8657 jobs). + +$ lake env lean AxiomCheck.lean # axiom audit of the two theorems +'BilliardsFormal.cot_lemma_L1' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.strictAntiOn_id_mul_cot' depends on axioms: [propext, Classical.choice, Quot.sound] diff --git a/problems/billiards-triangles/formal/lake-manifest.json b/problems/billiards-triangles/formal/lake-manifest.json new file mode 100644 index 0000000..5fcba8f --- /dev/null +++ b/problems/billiards-triangles/formal/lake-manifest.json @@ -0,0 +1,96 @@ +{"version": "1.2.0", + "packagesDir": ".lake/packages", + "packages": + [{"url": "https://github.com/leanprover-community/mathlib4", + "type": "git", + "subDir": null, + "scope": "", + "rev": "905b95818eb32af7874a58b427f50c1711a5e96c", + "name": "mathlib", + "manifestFile": "lake-manifest.json", + "inputRev": "v4.32.2", + "inherited": false, + "configFile": "lakefile.lean"}, + {"url": "https://github.com/leanprover-community/plausible", + "type": "git", + "subDir": null, + "scope": "leanprover-community", + "rev": "e12c1910fe855cbfc38803cd4e55543906d5fa62", + "name": "plausible", + "manifestFile": "lake-manifest.json", + "inputRev": "main", + "inherited": true, + "configFile": "lakefile.toml"}, + {"url": "https://github.com/leanprover-community/LeanSearchClient", + "type": "git", + "subDir": null, + "scope": "leanprover-community", + "rev": "c5d5b8fe6e5158def25cd28eb94e4141ad97c843", + "name": "LeanSearchClient", + "manifestFile": "lake-manifest.json", + "inputRev": "main", + "inherited": true, + "configFile": "lakefile.toml"}, + {"url": "https://github.com/leanprover-community/import-graph", + "type": "git", + "subDir": null, + "scope": "leanprover-community", + "rev": "7e9612bf0b9ee66db3cb5b9988a35afc706f5a12", + "name": "importGraph", + "manifestFile": "lake-manifest.json", + "inputRev": "main", + "inherited": true, + "configFile": "lakefile.toml"}, + {"url": "https://github.com/leanprover-community/ProofWidgets4", + "type": "git", + "subDir": null, + "scope": "leanprover-community", + "rev": "6e311e2a844da9b2cc3971187df2fe0066947b93", + "name": "proofwidgets", + "manifestFile": "lake-manifest.json", + "inputRev": "main", + "inherited": true, + "configFile": "lakefile.lean"}, + {"url": "https://github.com/leanprover-community/aesop", + "type": "git", + "subDir": null, + "scope": "leanprover-community", + "rev": "a7dbf0c63b694e47f425f3dcddbc0e178bb432d3", + "name": "aesop", + "manifestFile": "lake-manifest.json", + "inputRev": "master", + "inherited": true, + "configFile": "lakefile.toml"}, + {"url": "https://github.com/leanprover-community/quote4", + "type": "git", + "subDir": null, + "scope": "leanprover-community", + "rev": "38d591e778f100aec9762bb582f9c7f55f50e9dc", + "name": "Qq", + "manifestFile": "lake-manifest.json", + "inputRev": "master", + "inherited": true, + "configFile": "lakefile.toml"}, + {"url": "https://github.com/leanprover-community/batteries", + "type": "git", + "subDir": null, + "scope": "leanprover-community", + "rev": "023ce7d62a0531e22a5331e20b587817a80d49ff", + "name": "batteries", + "manifestFile": "lake-manifest.json", + "inputRev": "main", + "inherited": true, + "configFile": "lakefile.toml"}, + {"url": "https://github.com/leanprover/lean4-cli", + "type": "git", + "subDir": null, + "scope": "leanprover", + "rev": "88679d088c9720c27ebdf2ba4dafe17341747f94", + "name": "Cli", + "manifestFile": "lake-manifest.json", + "inputRev": "v4.32.0", + "inherited": true, + "configFile": "lakefile.toml"}], + "name": "BilliardsFormal", + "lakeDir": ".lake", + "fixedToolchain": false} diff --git a/problems/billiards-triangles/formal/lakefile.toml b/problems/billiards-triangles/formal/lakefile.toml new file mode 100644 index 0000000..b17d46b --- /dev/null +++ b/problems/billiards-triangles/formal/lakefile.toml @@ -0,0 +1,10 @@ +name = "BilliardsFormal" +defaultTargets = ["BilliardsFormal"] + +[[require]] +name = "mathlib" +git = "https://github.com/leanprover-community/mathlib4" +rev = "v4.32.2" + +[[lean_lib]] +name = "BilliardsFormal" diff --git a/problems/billiards-triangles/formal/lean-toolchain b/problems/billiards-triangles/formal/lean-toolchain new file mode 100644 index 0000000..0ec5999 --- /dev/null +++ b/problems/billiards-triangles/formal/lean-toolchain @@ -0,0 +1 @@ +leanprover/lean4:v4.32.2 diff --git a/problems/billiards-triangles/prior-art.json b/problems/billiards-triangles/prior-art.json index 1f950f1..22f7a65 100644 --- a/problems/billiards-triangles/prior-art.json +++ b/problems/billiards-triangles/prior-art.json @@ -244,6 +244,32 @@ ], "refuted_by": null, "gaps": [] + }, + { + "id": "009", + "file": "attempts/009-lean-pilot-cot-lemma.md", + "date": "2026-08-04", + "mode": "informed", + "status": "FORMALIZED", + "mechanism": [ + "machine-checked-formalization", + "symbolic-trig-closed-form" + ], + "one_line": "First Lean 4 certificate in the lab: Lemma L1 of 005 (c.cot(cs) strictly decreasing on (0,1] for fixed s in (0, pi/2]) proved in mathlib with zero sorries and only the three standard axioms, statement matching 005 hypothesis-by-hypothesis with no narrowing; independent statement review and independent rebuild still pending.", + "leak_terms": [ + "cot_lemma_L1", + "strictAntiOn_id_mul_cot", + "BilliardsFormal", + "no HasDerivAt lemma for Real.cot", + "first Lean certificate" + ], + "verifies": "005", + "range": "exactly the formal statement StrictAntiOn (fun c => c * Real.cot (c * s)) (Ioc 0 1) for all real s with 0 < s <= pi/2 (Lean 4.32.2, mathlib v4.32.2); Lemmas C and D themselves, and the I1-I4 Laurent block, are NOT formalized", + "gaps": [ + "independent-statement-review-pending (the FORMALIZE.md pass part 1 must be done by a non-formalizer agent)", + "independent-rebuild-pending (skeptic lake build on a different machine, FORMALIZE.md part 3)", + "laurent-block-I1-I4-not-attempted (queue item 16 second half)" + ] } ] } diff --git a/tests/test_records.py b/tests/test_records.py index ba4e101..be02919 100644 --- a/tests/test_records.py +++ b/tests/test_records.py @@ -100,7 +100,9 @@ def test_docs_do_not_point_at_missing_files(): ) broken = [] for doc in sorted(ROOT.rglob("*.md")): - if ".git" in doc.parts or "attempts" in doc.parts: + if ".git" in doc.parts or "attempts" in doc.parts or ".lake" in doc.parts: + # .lake/ holds fetched Lean dependencies (e.g. mathlib's own + # docs); git-ignored, not ours to lint. continue text = doc.read_text(encoding="utf-8", errors="ignore") for match in pattern.finditer(text): From dd3d86626be3e01de4ecefb6cd83c726d203a2fa Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 09:45:04 +0000 Subject: [PATCH 05/14] billiards-triangles 010: statement review of 009 - FORMALIZED pass complete Independent skeptic pass on the Lean pilot: statement fidelity vs 005's Lemma L1 confirmed hypothesis-by-hypothesis with no narrowing (Ioc, StrictAntiOn and Real.cot semantics read from mathlib source), vacuity/ cheat scan clean, root module confirmed to elaborate the theorem file, independent rebuild green with the reviewer's own axiom audit (propext, Classical.choice, Quot.sound only), all three L1 uses in 005 confirmed in-hypothesis. 009's two pending gaps closed in the index; 009's record untouched. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- .../attempts/010-statement-review-of-009.md | 302 ++++++++++++++++++ problems/billiards-triangles/prior-art.json | 22 +- 2 files changed, 322 insertions(+), 2 deletions(-) create mode 100644 problems/billiards-triangles/attempts/010-statement-review-of-009.md diff --git a/problems/billiards-triangles/attempts/010-statement-review-of-009.md b/problems/billiards-triangles/attempts/010-statement-review-of-009.md new file mode 100644 index 0000000..03dc2d1 --- /dev/null +++ b/problems/billiards-triangles/attempts/010-statement-review-of-009.md @@ -0,0 +1,302 @@ +# 010 — Statement review + independent rebuild of 009 (Lean pilot, Lemma L1) + +- **Problem:** billiards-triangles, `problems/billiards-triangles/PROBLEM.md` +- **Date:** 2026-08-04 +- **Mode:** informed (read `prior-art.json`, 009 in full, 005 §(ii)-B and its + L1 use sites, 008 §front matter for review conventions, `docs/FORMALIZE.md`, + the four `formal/` project files, `formal/lake-build.log`, and — for the + definition audit — the vendored mathlib source under + `formal/.lake/packages/mathlib`) +- **Type:** skeptic review of `009-lean-pilot-cot-lemma.md` (default stance: + REFUTE). This record is the independent completion of the three-part + `FORMALIZED` pass that `docs/FORMALIZE.md` defines: (1) statement review by + a non-formalizer, (2) clean-build confirmation, (3) toolchain/reproduction + check via my own `lake build`. I wrote none of the Lean code under review. +- **Outcome in one line:** 009 survives on every attacked front. The formal + statement is 005's Lemma L1 hypothesis-for-hypothesis with no narrowing; + the source is cheat-free and non-vacuous; my own fresh `lake build` + (project build output deleted first) is green and my own axiom audit shows + exactly `[propext, Classical.choice, Quot.sound]`; all three of 005's uses + of L1 instantiate inside the formalized hypotheses. Two cosmetic nits, no + corrections. The FORMALIZED pass on 005's Lemma L1 is now **complete**. +- **Tools:** grep over `formal/` and the vendored mathlib (definition audit, + cheat scan); elan 4.2.3 / Lean `leanprover/lean4:v4.32.2` via + `$HOME/.elan/bin` for the rebuild (`rm -rf .lake/build && lake build`, + deterministic, CotMonotone rebuilt from source in 27 s) and the axiom audit + (a temporary `SkepticAxiomCheck.lean` in `formal/`, deleted after the run; + exact commands and output below). No new committed code. +- **Sources:** repo + the vendored mathlib source only; no external papers. + +No file of 009 (or any prior attempt) was modified. + +## Claims attacked + +The attack list, (a)–(f) as assigned plus (g)–(h) of my own: + +1. **(a) Statement fidelity.** That + `theorem cot_lemma_L1 {s : ℝ} (hs0 : 0 < s) (hs : s ≤ π / 2) : + StrictAntiOn (fun c : ℝ => c * Real.cot (c * s)) (Set.Ioc 0 1)` + says exactly what 005 §(ii)-B's L1 says ("For fixed s ∈ (0, π/2], the map + c ↦ c·cot(cs) is strictly decreasing on (0, 1]") — variable roles, + interval endpoints and open/closed-ness, strictness direction, s-range. +2. **(b) Definition audit.** That mathlib's `Real.cot`, `Set.Ioc`, + `StrictAntiOn` mean what 009's mapping table asserts, checked against the + vendored mathlib *source* (not docs), and that no junk value of `Real.cot` + at a pole of sin is relied on inside the claimed domain. +3. **(c) Vacuity / cheat scan.** Every line of `CotMonotone.lean` and + `BilliardsFormal.lean` read for `sorry`/`admit`/`native_decide`/ + `axiom`/`unsafe`/`partial`/`@[implemented_by]`/custom `notation`/`macro`/ + shadowed `cot`/`π`; that the hypotheses are satisfiable and the domain + nonempty; that the root module actually imports the theorem file (a + `lake build` that never elaborates it would pass trivially). +4. **(d) Independent rebuild + axiom audit** (FORMALIZE.md parts 2–3): my own + `lake build` after deleting the project's own build output, and my own + `#print axioms` run, not trusting `lake-build.log`. +5. **(e) Record audit.** Whether 009's prose or its prior-art.json entry + claims more than the certificate shows (in particular anything about + "retiring Lemmas C and D", which are informal consumers 009 did not + formalize), and whether `range` is scoped honestly. +6. **(f) Usage check.** That 005 actually leans on L1 *as formalized* — each + use site instantiates (s, c-pair) inside `0 < s ≤ π/2`, `c ∈ (0, 1]` — + so the certificate covers a statement 005 really consumes, not a variant. +7. **(g, own) Proof-mathematics hand-check.** The quotient-rule expression, + the Pythagorean simplification, and the final inequality re-derived by + hand (the kernel guarantees the formal steps; this checks that what the + kernel accepted is the argument 005 and 009 describe, i.e. no smuggled + alternative proof with different domain needs). +8. **(h, own) Toolchain-pin consistency + 009's tooling claims.** That + `lean-toolchain`, `lakefile.toml`, `lake-manifest.json`, and the log's + `lean --version` line all pin the same versions 009 reports, and that + 009's "mathlib v4.32.2 has no derivative API for `Real.cot`" is true. + +## Refutations found + +**None.** No hypothesis mismatch, no narrowing, no cheat, no axiom beyond +the standard three, no vacuity, no overclaim in the record. Two cosmetic +nits, recorded here because a skeptic pass should say what it noticed +(neither is a correction; nothing downstream changes): + +1. `formal/lake-build.log` line 8 shows the formalizer's axiom audit run as + `lake env lean AxiomCheck.lean`, but no `AxiomCheck.lean` is committed — + it was evidently a scratch file. Reproduction therefore requires + recreating it (two lines; contents inferable from the log). I re-ran the + audit myself from a fresh scratch file, so the certificate does not rest + on that log line. +2. The module docstring of `CotMonotone.lean` (lines 9–11) says the lemma + "retires Lemmas C and D". As a description of 005's usage this is + accurate, but read in isolation from the .lean file it could be taken as + a claim about what is *formalized*. The record 009 itself scopes this + correctly and explicitly ("Lemmas C and D … remain at `VERIFIED` …, not + `FORMALIZED`"), so this is a docstring-tone nit only. + +## Claims that survive + +All of them. Item by item, with what I did to try to break each: + +### (a) Statement fidelity — CONFIRMED, no deltas + +Checked against 005 §(ii)-B lines 157–158 verbatim: + +| 005's L1 | formal | verdict | +|---|---|---| +| "fixed s ∈ (0, π/2]" | `{s : ℝ} (hs0 : 0 < s) (hs : s ≤ π / 2)` — universally quantified over exactly (0, π/2] | match ("fixed s" = ∀ s; two inequalities = the half-open interval, closed at π/2 as 005 needs for Lemma D's endpoint s = bx = π/2) | +| variable roles: c is the moving variable, s the parameter | the lambda binds `c`; `s` is fixed outside `StrictAntiOn` | match — no c/s swap | +| "the map c ↦ c·cot(cs)" | `fun c : ℝ => c * Real.cot (c * s)` | match (radians: `Real.cot`; 005 says "radians throughout this subsection") | +| "strictly decreasing on (0, 1]" | `StrictAntiOn … (Set.Ioc 0 1)` | match — see (b) for both definitions read from source | + +No hypothesis is stronger than 005's, no case dropped, no strictness +weakened; the interval is (0, 1] closed at 1 exactly as 005 needs (c = 1 is +one side of all three of 005's comparisons). 009's mapping table is signed +off unchanged. + +### (b) Definition audit — CONFIRMED from vendored mathlib source + +Read at `formal/.lake/packages/mathlib` (rev `905b958…`, tag v4.32.2): + +- `Real.cot` (`Mathlib/Analysis/Complex/Trigonometric.lean:88`) is + `(Complex.cot x).re` with `Complex.cot z = cos z / sin z` (line 44), and + `Real.cot_eq_cos_div_sin : cot x = cos x / sin x` (line 641). Standard + convention, radian. +- Poles: Lean's `x / 0 = 0`, so `Real.cot` is total with junk value 0 where + sin = 0. On the claimed domain nothing touches a pole: c ∈ (0,1], + s ∈ (0,π/2] gives cs ∈ (0, π/2] ⊂ (0, π), where the proof itself + establishes sin > 0 (`Real.sin_pos_of_pos_of_lt_pi`, line 51–52 of + `CotMonotone.lean`). The one global step, the `funext` rewrite + `cot u = cos u / sin u`, is a definitional identity valid at every real + including poles (both sides are the same junk there), and monotonicity is + only ever asserted on the pole-free `Ioc`. No junk value is load-bearing. +- `StrictAntiOn f s` (`Mathlib/Order/Monotone/Defs.lean:106`) is + `∀ ⦃a⦄, a ∈ s → ∀ ⦃b⦄, b ∈ s → a < b → f b < f a` — i.e. strictly + decreasing on s, the direction 005 claims. (Attack considered: a + StrictAnti/StrictAntiOn or Anti/AntiOn confusion, or ≤ in the conclusion. + Neither is present.) +- `Set.Ioc a b` (`Mathlib/Order/Interval/Set/Defs.lean:63`) is + `{x | a < x ∧ x ≤ b}` = (a, b]. So `Ioc 0 1` = (0, 1] and + `Ioc 0 (π/2)` = (0, π/2], as 009 states. + +### (c) Vacuity / cheat scan — CLEAN + +- `grep -inE "sorry|admit|native_decide|axiom|unsafe|partial|implemented_by|notation|macro|elab"` + over both project .lean files: the only hit is the word "sorry" inside + `CotMonotone.lean`'s doc comment. No `local notation`, no custom `cot` or + `π`: the file's `open Real Set` resolves `π` to `Real.pi` and every `cot` + is written qualified as `Real.cot`. +- Vacuity: the hypotheses are jointly satisfiable (e.g. s = 1) and + `Set.Ioc 0 1` is nonempty, so the theorem has content; `hs : s ≤ π / 2` + cannot empty anything. +- Elaboration reachability: `BilliardsFormal.lean` is exactly + `import BilliardsFormal.CotMonotone`; `lakefile.toml`'s default target is + the `BilliardsFormal` lib rooted there; my own build log (below) shows + both `BilliardsFormal.CotMonotone` and `BilliardsFormal` compiled. These + two files are the only .lean sources outside `.lake/` — nothing else to + audit. + +### (d) Independent rebuild + axiom audit — GREEN (FORMALIZE.md parts 2–3) + +Exact commands, from `problems/billiards-triangles/formal/`, PATH prefixed +with `$HOME/.elan/bin` (elan 4.2.3): + +``` +$ rm -rf .lake/build # project's own output only; .lake/packages kept +$ lake build +✔ [8655/8657] Built BilliardsFormal.CotMonotone (27s) +✔ [8656/8657] Built BilliardsFormal (7.1s) +Build completed successfully (8657 jobs). + +$ printf 'import BilliardsFormal\n#print axioms BilliardsFormal.cot_lemma_L1\n#print axioms BilliardsFormal.strictAntiOn_id_mul_cot\n' > SkepticAxiomCheck.lean +$ lake env lean SkepticAxiomCheck.lean +'BilliardsFormal.cot_lemma_L1' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.strictAntiOn_id_mul_cot' depends on axioms: [propext, Classical.choice, Quot.sound] +$ rm SkepticAxiomCheck.lean # scratch file, not committed +``` + +Exactly the three standard axioms, no `sorryAx`, no project-local axiom. +Scope of independence, stated honestly: this is a different agent and a +fresh container session from 009's, the project's own build products were +deleted and re-elaborated from the committed source by the pinned toolchain, +but the mathlib *dependency* oleans came from the same on-disk +`.lake/packages` cache (per the review brief: mathlib was not re-downloaded +or rebuilt). The trust base is therefore the pinned mathlib cache + Lean +kernel, which is the trust base every mathlib user accepts. + +### (e) Record audit — HONEST + +- 009 flags its own mapping table `PENDING INDEPENDENT STATEMENT REVIEW`, + states outright that Lemmas C and D "remain at `VERIFIED` …, not + `FORMALIZED`", and its final section says the pass is incomplete until an + independent rebuild + statement review — i.e. it under-claims, which is + the correct direction. (This record is that completion.) +- prior-art.json 009 entry: `range` names exactly the formal statement and + says C/D and I1–I4 are NOT formalized; `gaps` lists + independent-statement-review-pending and independent-rebuild-pending. + Honest; both gaps are closed by this record and removed from 009's index + entry (index entries are living state; 009's .md is untouched). +- Numbers spot-checked against the log: "compiles in ~8 s" vs log's 7.5 s, + 8657 jobs, timestamps — consistent. +- The one place a hasty reader could over-read is the .lean docstring's + "retires Lemmas C and D" (nit 2 above); the record itself never makes + that mistake. + +### (f) Usage check — 005 consumes L1 exactly as formalized + +All three use sites re-read in 005 §(ii)-B (lines 174–177, 187): + +1. Lemma C, first bracket: s = v ∈ (0, v₀], v₀ = π/(2(b+1)) ≤ π/4 < π/2; + compares c = b/a vs c = 1, with 1 ≤ b ≤ a so b/a ∈ (0, 1]. In-domain. + (005 writes the bracket "≤ 0, equality iff a = b": when b/a = 1 the two + c's coincide and the bracket is identically 0 — the strict formal + statement is exactly what the b/a < 1 case needs, nothing weaker + suffices, nothing stronger is used.) +2. Lemma C, second bracket: s = π/2 − v ∈ (0, π/2); compares c = 1/a ≤ 1/2 + vs c = 1 (a ≥ 2). In-domain, strict as 005 claims. +3. Lemma D: s = bx with x ∈ (0, π/(2b)], so s ∈ (0, π/2] — this is the use + that needs the closed endpoint s = π/2; compares c = 1/b vs c = 1 + (b ≥ 2). In-domain. + +No other form of L1 (different constants, different interval) appears +anywhere in 005; lines 379 and 447–448 refer back to the same statement. The +formalized statement is the used statement; the certificate is not vacuous. + +### (g) Proof-mathematics hand-check — the accepted proof is 005's argument + +- `hasDerivAt_id_mul_cot`: quotient rule on (u·cos u)/sin u with + f′ = cos x − x·sin x, g′ = cos x — the stated derivative expression is the + textbook (f′g − fg′)/g². +- The `linear_combination`-justified simplification: + (cos x − x sin x)·sin x − x·cos²x = sin x cos x − x·(sin²x + cos²x) + = sin x cos x − x. Hand-checked; matches 005's + [sin(2cs) − 2cs]/(2 sin²(cs)) after the double-angle rewrite, as 009's + "proof-shape delta" paragraph claims. +- Negativity: sin x·cos x ≤ sin x (sin ≥ 0, cos ≤ 1) and + `Real.sin_lt : 0 < x → sin x < x` (read from source, + `Mathlib/Analysis/SpecialFunctions/Trigonometric/Bounds.lean:42`) — the + same "sin x < x" driver 005 cites. +- The u = c·s transfer in `cot_lemma_L1`: c₁ < c₂ gives c₁s < c₂s + (`mul_lt_mul_of_pos_right`, s > 0), the core theorem gives + (c₂s)cot(c₂s) < (c₁s)cot(c₁s), and dividing by s > 0 + (`lt_of_mul_lt_mul_left`) gives the goal. Direction-checked by hand: this + really is f(c₂) < f(c₁), the StrictAntiOn conclusion, not its reverse. + +### (h) Toolchain pins — CONSISTENT; tooling claims TRUE + +`lean-toolchain` = `leanprover/lean4:v4.32.2`; `lakefile.toml` requires +mathlib at `rev = "v4.32.2"`; `lake-manifest.json` locks mathlib at +`905b95818eb32af7874a58b427f50c1711a5e96c` with `inputRev v4.32.2`; the +log's `lean --version` says 4.32.2. All four agree with 009's front matter. +009's claim that this mathlib has no derivative API for `Real.cot`: +confirmed by grep over the vendored `Mathlib/Analysis/` — no +`hasDerivAt_cot`/`deriv_cot` for `Real` (only `arccot`/`coth` hits), so the +inline quotient-rule lemma was genuinely necessary. + +### Verdict on the three-part FORMALIZED pass (docs/FORMALIZE.md) + +1. **Statement review** (the point): DONE by this record, by a + non-formalizer, hypothesis by hypothesis — no narrowing, no deltas. +2. **Clean build**: CONFIRMED — zero `sorry`, axioms exactly + `[propext, Classical.choice, Quot.sound]`, full source in the repo, + verified by my own audit, not the formalizer's log. +3. **Recorded toolchain**: CONFIRMED — pins consistent across all four + files, and my own independent `lake build` (fresh session, project build + output deleted) is green. + +**The FORMALIZED status of 009 (= the machine-checked certificate for 005's +Lemma L1) is complete.** Scope unchanged from 009's own statement: exactly +`cot_lemma_L1` and its core `strictAntiOn_id_mul_cot`; Lemmas C, D and the +I1–I4 Laurent block remain informal (`VERIFIED` via 008). + +What is *not* claimed by this review: no re-verification of 005's Lemmas +C/D or anything else in 005 (008 did that); no audit of mathlib itself +beyond reading the definitions used; no rebuild of mathlib from source; no +check of Lean-kernel soundness. + +## Leads generated + +Unchanged from 009's leads 2–4 (Lemma D next, then I1–I4, then Lemma C) — +this review adds no new ones and found nothing that spawns any. One +process lead: + +1. **Commit the axiom-audit scratch file pattern.** Add the two-line + `AxiomCheck.lean` to `formal/` (it is part of the reproduction recipe + the log references but the repo does not contain). Definite outcome: + `lake env lean AxiomCheck.lean` works from a clean checkout with no + file-recreation step, or the file is deliberately left out and the log + line should carry the file's contents instead. + +## References + +- `problems/billiards-triangles/attempts/009-lean-pilot-cot-lemma.md` (the + record under review). +- `problems/billiards-triangles/attempts/005-complete-death-law-theorem.md` + §(ii)-B (Lemma L1 statement, lines 157–161; use sites lines 174–177, 187). +- `problems/billiards-triangles/attempts/008-skeptic-review-of-005.md` + (prior hand verification of L1; review-shape conventions). +- `docs/FORMALIZE.md` (the three-part pass this record completes). +- `problems/billiards-triangles/formal/` — `BilliardsFormal/CotMonotone.lean`, + `BilliardsFormal.lean`, `lakefile.toml`, `lean-toolchain`, + `lake-manifest.json`, `lake-build.log`. +- mathlib4 tag v4.32.2 (rev `905b95818eb32af7874a58b427f50c1711a5e96c`), + read from the vendored copy under `formal/.lake/packages/mathlib`: + `Analysis/Complex/Trigonometric.lean` (cot), `Order/Monotone/Defs.lean` + (StrictAntiOn), `Order/Interval/Set/Defs.lean` (Ioc), + `Analysis/SpecialFunctions/Trigonometric/Bounds.lean` (sin_lt). No + papers; no machine transcriptions. diff --git a/problems/billiards-triangles/prior-art.json b/problems/billiards-triangles/prior-art.json index 22f7a65..ec3431b 100644 --- a/problems/billiards-triangles/prior-art.json +++ b/problems/billiards-triangles/prior-art.json @@ -266,10 +266,28 @@ "verifies": "005", "range": "exactly the formal statement StrictAntiOn (fun c => c * Real.cot (c * s)) (Ioc 0 1) for all real s with 0 < s <= pi/2 (Lean 4.32.2, mathlib v4.32.2); Lemmas C and D themselves, and the I1-I4 Laurent block, are NOT formalized", "gaps": [ - "independent-statement-review-pending (the FORMALIZE.md pass part 1 must be done by a non-formalizer agent)", - "independent-rebuild-pending (skeptic lake build on a different machine, FORMALIZE.md part 3)", "laurent-block-I1-I4-not-attempted (queue item 16 second half)" ] + }, + { + "id": "010", + "file": "attempts/010-statement-review-of-009.md", + "date": "2026-08-04", + "mode": "informed", + "status": "VERIFIED", + "mechanism": [ + "adversarial-verification", + "machine-checked-formalization" + ], + "one_line": "Independent completion of the FORMALIZE.md pass on 009: non-formalizer statement review finds cot_lemma_L1 matches 005's Lemma L1 hypothesis-for-hypothesis with no narrowing (StrictAntiOn/Ioc/Real.cot read from vendored mathlib source, no junk-value reliance, cheat scan clean), own fresh lake build after deleting the project build output is green with axioms exactly [propext, Classical.choice, Quot.sound], and all three of 005's L1 use sites instantiate inside the formalized hypotheses; zero corrections, two cosmetic nits.", + "leak_terms": [ + "SkepticAxiomCheck", + "statement review signed off", + "AxiomCheck.lean not committed" + ], + "verifies": "009", + "range": "statement fidelity vs 005 section (ii)-B checked hypothesis by hypothesis; definitions Real.cot, StrictAntiOn, Set.Ioc read from mathlib rev 905b958 source; cheat scan over both project .lean files; rebuild = rm -rf .lake/build + lake build (8657 jobs, green) in a fresh container session with the pinned toolchain, mathlib oleans from the existing pinned cache (not re-downloaded); own axiom audit of both theorems; usage check of all three L1 call sites in 005. NOT audited: mathlib internals beyond the definitions used, Lean kernel, Lemmas C/D themselves", + "gaps": [] } ] } From a0bfd6ed660fb4ab8ced4158ba491c87a4667cd8 Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 09:46:22 +0000 Subject: [PATCH 06/14] Ledger: close the 2026-08-04 Astra follow-through cycle MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Ripple scan of the ten Astra results: zero hits across the portfolio; both flagged bites (Ehrhart vs mahler-4d, extremal/Ramsey vs erdos-gyarfas) recorded as reasoned misses, one conditional analysis-lens seed kept as SPECULATION in insights, two scan hazards recorded (name collision, Erdős numbering conflict). Formalization pilot closed: L1 certificate entry added to verified results, queue items 15/16 retired, new item 15 targets the I1-I4 Laurent block. GUIDANCE.md directives updated to completed state; ops notes on the Lean toolchain recorded in insights. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- GUIDANCE.md | 25 ++++++++++++-------- STATUS.md | 67 ++++++++++++++++++++++++++++++++++++++++++++++------- 2 files changed, 73 insertions(+), 19 deletions(-) diff --git a/GUIDANCE.md b/GUIDANCE.md index b39c0c5..735a583 100644 --- a/GUIDANCE.md +++ b/GUIDANCE.md @@ -15,22 +15,27 @@ for a reported ~$2,000 of tokens. None are peer-reviewed yet; treat the individual results as reported, not settled. The *shape* of the release is what matters here, and three things follow for this lab: -1. **Run the ripple scan first.** Before the next cycle pulls anything else - from the queue, run `docs/RIPPLE.md` on the Astra results (queue item 15). - The Ehrhart-volume and extremal-graph/Ramsey entries are the plausible - bites on mahler-4d and erdos-gyarfas; the scan decides, not this note. +1. **Ripple scan: done (2026-08-04), zero hits.** The scan ran the same day + this guidance was written; no gap unblocked, no route foreclosed — the + full reasoning is in the `STATUS.md` insights entry. The one live + residue: when a manuscript mirror becomes reachable, spend the hour + checking whether the Astra Ehrhart proof is complex-analytic; only in + that case, seed an analysis-lens `/ideate` sweep on mahler-4d. 2. **The novelty gate is now part of the bar.** The October 2025 "GPT-5 solved ten Erdős problems" collapse was a labeling failure, not a derivation failure: correct arguments, already in the literature, claimed as new. Our blind mode *wants* rediscoveries — but recorded as rediscoveries. The skeptic pass now includes a literature check on any claimed-new result (`docs/CYCLE.md` step 4, `CONTRIBUTING.md` bar item 7). -3. **Formalization is the new top verification rung.** For skeptic-confirmed, - load-bearing, proof-shaped steps, a kernel-checked Lean certificate is the - one thing stronger than a second independent derivation — it is what made - Astra's claims land after the October episode. `docs/FORMALIZE.md` - defines the lane and the `FORMALIZED` status; queue item 16 is the pilot. - Expensive; use deliberately. +3. **Formalization is the new top verification rung — and the lane is now + operational.** For skeptic-confirmed, load-bearing, proof-shaped steps, a + kernel-checked Lean certificate is the one thing stronger than a second + independent derivation — it is what made Astra's claims land after the + October episode. `docs/FORMALIZE.md` defines the lane and the + `FORMALIZED` status; the pilot completed 2026-08-04 (billiards 005's + Lemma L1, attempts 009+010, full three-part pass). Next certificate + target is the I1–I4 Laurent block (queue item 15). Expensive; use + deliberately. **Portfolio note.** Astra's target list — decade-open but specialist-tractable problems whose deliverables are checkable objects diff --git a/STATUS.md b/STATUS.md index 1ffe79e..994d36c 100644 --- a/STATUS.md +++ b/STATUS.md @@ -11,12 +11,16 @@ out, and no cron trigger is active. Nothing is half-finished: every line is written up, and the queue below is a list of starting points rather than abandoned work. -**Process update (2026-08-04), prompted by OpenAI's Astra announcement:** -a literature-novelty check is now part of the skeptic pass, a formalization -lane (`docs/FORMALIZE.md`, status `FORMALIZED`) sits above `VERIFIED` for -load-bearing proof steps, and `GUIDANCE.md` directs the next cycle to start -with a ripple scan of the Astra results (queue items 15–16). No mathematical -state changed. +**2026-08-04 cycle (Astra follow-through):** a literature-novelty check is +now part of the skeptic pass and a formalization lane (`docs/FORMALIZE.md`, +status `FORMALIZED`) sits above `VERIFIED` for load-bearing proof steps. +Both queue items the new guidance created are done: the ripple scan of the +ten Astra results ran with **zero hits** across the portfolio (see +insights — both flagged bites are reasoned misses), and the formalization +pilot landed: billiards 005's Lemma L1 is the lab's first machine-checked +result (Lean 4 + mathlib, attempts 009+010, statement review and +independent rebuild complete). No mathematical state changed beyond the +new certificate. To pick something up, take an item from the attempt queue and follow `docs/CYCLE.md` — either by hand, one line at a time, or by restarting an @@ -67,7 +71,7 @@ attempts (queue; run blind). | Lonely runner | k=8 done | medium | next: k=9 scan (k=8 likely settled by Rosenfeld preprint) | | Graceful trees | census done (n≤14) | low | possible next: mine the symmetric-spider seed; lobster verification at larger n | | Collatz | queued | low (long shot) | failed-approach taxonomy; cycle-bound frontier | -| Triangular billiards | death law CLOSED both sides: parametric necessity all (a,b), death = γ_d exactly for 20 members (005/008); 135° stall dissolved, birth law + exact-135 certificates (006/007) | high | next: parametric sufficiency + birth-law theorem (queue 11); coverage conjecture + sampler blind spot (queue 12) | +| Triangular billiards | death law CLOSED both sides: parametric necessity all (a,b), death = γ_d exactly for 20 members (005/008); 135° stall dissolved, birth law + exact-135 certificates (006/007) | high | next: parametric sufficiency + birth-law theorem (queue 11); coverage conjecture + sampler blind spot (queue 12); Lean lane open — L1 FORMALIZED (009/010), I1–I4 Laurent block is the next certificate target | | Mahler in ℝ⁴ | census done blind (skeptic-confirmed) | medium | next: close k=12–20 (falsifiable: no proper mask with P<11); run the same pipeline on {0,±1}³ for the n=3 spectrum comparison | | Crouzeix | onboarded, no attempts | medium | harness ready, certification risk retired; first attempt is the dim-3 landscape (run blind) | @@ -87,8 +91,7 @@ attempts (queue; run blind). 12. [billiards-triangles] **The coverage conjecture, and the sampler blind spot** (from 006/007, 2026-07-31; absorbs the old pinch-gap item — its motivating gap [135.000°, 135.049°] is CLOSED, W(4,3) is certified alive inside it): 006 reduced "every obtuse angle has an alive W member" to an elementary Diophantine statement (unproven; float-checked at 157 + 25 arcs over 90.5°–165°, zero failures). Prove it, using the birth law as a labelled input where needed. Note the certificates so far are POINTWISE (007's C2): window-interval continuity on sub-arcs is float + SPECULATION law only, and per-triangle coverage of a whole arc is a different (open) question — the windows are x-slivers at the corners. Separately falsifiable (007 lead): every sampler in use accumulates only at the 90/j window edges, so an interior-pinch alive window would hide from ALL current designs — build one targeted interior-accumulation test before trusting any negative screen again. 13. [mahler-4d] Close the {0,±1}⁴ universe: k = 12–20 pairs (~30M orbits at k=12, improper fraction already 77% at k=9). Falsifiable: no proper mask with k ≥ 12 has P < 11. Needs the improper-detection shortcut or a streaming canonicalizer; see 001 lead 1. Cheap side quest, same pipeline: the {0,±1}³ census for the n=3 spectrum comparison (13 pairs, trivial) — does the non-Hanner gap grow or shrink with n? 14. [billiards-triangles] Coverage self-test: re-derive the acute and right-triangle cases as a scoped attempt record. Low value now that the harness self-test covers Fagnano and the orthic geometry and 001 mapped the obtuse side — take it only if something turns up that the certificate machinery cannot express. -15. [cross-problem] **Ripple scan the Astra results** (GUIDANCE 2026-08-04 pulls this to the front of the next cycle). Per `docs/RIPPLE.md`: characterize each of the ten 2026-08-02 OpenAI results (scope, mechanism tags, what the argument needs to apply), then scan the portfolio. Plausible bites to check first, without prejudging: the Ehrhart volume-conjecture proof against mahler-4d (both live in lattice-polytope volume land — does its machinery say anything about volume products, or vice versa?), and the extremal-graph-theory / multicolor-Ramsey entries against erdos-gyarfas (cycle-spectrum arguments share the field lens). Verify each result's actual statement from the published manuscripts/Lean certificates, not press coverage — these are unrefereed, so scope-check harder than usual, and treat any transfer hypothesis as SPECULATION. Expected outcome: recorded misses; a hit goes to the top of that problem's queue; a scan that changes the picture becomes a MAP attempt. -16. [billiards-triangles] **Formalization pilot** (`docs/FORMALIZE.md`): produce the lab's first Lean 4 certificate for the c·cot(cs) monotonicity lemma (elementary, retires Lemmas C and D), then attempt the Laurent-identity block I1–I4 (already one exact polynomial check — but its geometry bridge is exactly what the statement-review step exists for). Record as a review-shape attempt with `status: FORMALIZED`; toolchain, mathlib pin, and `lake build` output in the record; independent statement review by a non-formalizer agent is part of the pass. A failed pilot is a full deliverable if it pins whether the obstruction is mathematical or tooling. +15. [billiards-triangles] **Second certificate: the I1–I4 Laurent-identity block** (`docs/FORMALIZE.md`; follows the completed L1 pilot, 009/010). The identities are already one exact polynomial check after adjoining e^{iaα}, e^{ibβ} as formal variables — formalizing the polynomial identity is the tractable half, and the geometry bridge (unfolding = closed form) is exactly what the statement-review step exists to interrogate; 010 kept this gap open deliberately. Harder than the pilot: needs Laurent-polynomial infrastructure in mathlib, and the statement must be pinned to what 005 actually proves (formal identity + ring-homomorphism specialization), not to the geometric claim it feeds. ## Verified results @@ -321,8 +324,54 @@ attempts (queue; run blind). was a relabeling. Corrections: length-30 universe is 566 words (not 811); gap certificates are pointwise, not "alive throughout". +- **[billiards-triangles] Lemma L1 FORMALIZED — the lab's first + machine-checked certificate (skeptic-confirmed)** (2026-08-04, attempts + 009+010): 005's L1 (for fixed s ∈ (0, π/2], c ↦ c·cot(cs) is strictly + decreasing on (0, 1] — the lemma that retires Lemmas C and D) proved in + Lean 4 (mathlib v4.32.2), zero sorries, axioms exactly {propext, + Classical.choice, Quot.sound}. Independent statement review confirmed + fidelity to 005 hypothesis-by-hypothesis with no narrowing (Ioc, + StrictAntiOn, and Real.cot semantics read from mathlib source), the + cheat scan is clean, the root module provably elaborates the theorem + file, and the reviewer's own rebuild + axiom audit went green. Scope: + exactly the formal statement — L1 only; Lemmas C/D themselves and the + I1–I4 Laurent block remain VERIFIED, not FORMALIZED (queue 15). Source + in problems/billiards-triangles/formal/. + ## Insights / cross-problem notes +- Ripple scan of the Astra results ran (2026-08-04, informed): + all ten results characterized (press-transcribed only — openai.com, the + certificate repo, and erdosproblems.com were unreachable, so every scope + statement is [L]-grade and must be re-checked against a manuscript + before any transfer work) and scanned against all ten problems. **No + purchase anywhere**: no recorded gap unblocked, no queued route + foreclosed. Both queue-flagged bites are reasoned misses — Ehrhart vs + mahler-4d fails on direction (upper vs lower bound), symmetry (Ehrhart's + content is the non-symmetric case; the symmetric case is classical + Minkowski), and hypotheses (Mahler has no lattice-point constraint); + extremal/Ramsey vs erdos-gyarfas fails on regime (Turán/dense and + complete-graph coloring vs bounded-degree cycle forcing — a cubic graph + sits below every Turán threshold in play). One conditional seed, + SPECULATION, deliberately not queued: if the Astra Ehrhart proof turns + out complex-analytic (Berman–Berndtsson lineage — the toolbox + historically shared with Bourgain–Milman), it becomes the seed for an + analysis-lens /ideate sweep on mahler-4d; check the method (one hour) + when a manuscript mirror is reachable. Two hazards recorded for future + scans: the "Erdős–Gyárfás generalized Ramsey (p,q)-coloring" problem in + Astra-adjacent press is NOT our cycle conjecture; and press sources + conflict on which of Erdős #146/#180 is compactness vs degeneracy — do + not cite that numbering from this scan. +- First FORMALIZE.md pass ran end-to-end (2026-08-04, billiards 009+010), + ops notes: the toolchain is viable in-session (elan, lean4 v4.32.2, + prebuilt mathlib via `lake exe cache get`, ~7.4 GB — never build mathlib + from source; `formal/.lake/` must be git-ignored, and the docs lint now + skips vendored deps). mathlib v4.32.2 has Real.cot but no derivative API + for it — an inline quotient rule is three lines, not an obstruction. The + statement-review step earned its keep procedurally: the semantic + verdicts (interval endpoints, StrictAntiOn direction, cot convention) + were read from mathlib source rather than assumed — exactly the + informal→formal bridge risk the lane exists to cover. - External event + process change (2026-08-04): OpenAI announced (2026-08-02) ten results on decade-plus-open problems (non-sofic groups, Connes rigidity counterexample, Ehrhart volume conjecture, three Erdős-catalog problems From 6bbbb675824f0caa85369b549adfa056c0143521 Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 10:20:30 +0000 Subject: [PATCH 07/14] billiards-triangles 011: Laurent-block certificate - tiers A+B+C green, review pending Second certificate of the FORMALIZE.md lane: 005's closed-form composition, the glide/translation facts, and identities I1-I4/D1-D2 proved in Lean 4 over an arbitrary field with (a,b) carried as actual powers of universally quantified generators; specialization corollaries included. 23 theorems, zero sorries, axioms propext/Classical.choice/ Quot.sound only, toolchain pins unchanged. Two findings recorded: the block is homogeneous in i (the i^2 = -1 hypothesis is never needed), and no Laurent-ring API is required (field_simp + ring_nf suffice). Geometry bridge explicitly out of scope; statement review and independent rebuild pending per the 009->010 precedent. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- .../attempts/011-lean-laurent-block.md | 272 ++++++++ .../formal/AxiomCheck.lean | 40 ++ .../formal/BilliardsFormal.lean | 1 + .../formal/BilliardsFormal/LaurentBlock.lean | 423 ++++++++++++ .../billiards-triangles/formal/lake-build.log | 33 + problems/billiards-triangles/prior-art.json | 611 +++++++++--------- 6 files changed, 1088 insertions(+), 292 deletions(-) create mode 100644 problems/billiards-triangles/attempts/011-lean-laurent-block.md create mode 100644 problems/billiards-triangles/formal/AxiomCheck.lean create mode 100644 problems/billiards-triangles/formal/BilliardsFormal/LaurentBlock.lean diff --git a/problems/billiards-triangles/attempts/011-lean-laurent-block.md b/problems/billiards-triangles/attempts/011-lean-laurent-block.md new file mode 100644 index 0000000..dd41792 --- /dev/null +++ b/problems/billiards-triangles/attempts/011-lean-laurent-block.md @@ -0,0 +1,272 @@ +# 011 — Lean 4 certificate for 005's Laurent-identity block (I1–I4, glide facts, closed-form composition) + +- **Problem:** billiards-triangles, `problems/billiards-triangles/PROBLEM.md` +- **Date:** 2026-08-04 +- **Mode:** informed (read `prior-art.json`, 005 in full, 008 §§1–2, + `docs/FORMALIZE.md`, `STATUS.md` queue item 15 and the 2026-07-31 + formal-parameter-specialization insight, 009 in full, the 009/010 index + entries, and `explore/plaw_general.py`'s ring layer — the proof object + being formalized) +- **Type:** formalization — review-shape attempt against + `005-complete-death-law-theorem.md` (verifies 005 §(ii)-A Steps 2–4: the + closed-form composition of the half word and the identity block I1–I4, + D1, D2 + glide facts, with (a,b) as formal parameters, now + kernel-checked). Second run of the `docs/FORMALIZE.md` lane, following + the completed L1 pilot (009/010); queue item 15. +- **Tools:** the existing Lean project at + `problems/billiards-triangles/formal/` — toolchain UNCHANGED from + 009/010: `leanprover/lean4:v4.32.2`, mathlib tag `v4.32.2` + (rev `905b95818eb32af7874a58b427f50c1711a5e96c`), prebuilt cache reused, + nothing re-downloaded. New source: + `formal/BilliardsFormal/LaurentBlock.lean` (~420 lines, 23 theorems), + imported from `formal/BilliardsFormal.lean`; `formal/AxiomCheck.lean` + (the audit driver, now on disk — 010's nit). `CotMonotone.lean` + untouched. Build is deterministic; the new module elaborates in ~24 s; + build + axiom audit appended to `formal/lake-build.log`. A throwaway + numeric transcription check (random nonzero complex points, session + scratchpad, not kept) was used to pin the encoding before proving; the + Lean proofs supersede it. +- **Sources:** none external beyond mathlib. Repo: 005 (the record + verified), 008 (skeptic confirmation of the same block, incl. the pair + "off-torus" algebra this encoding follows), 009/010 (lane precedent), + `docs/FORMALIZE.md`. + +**Forbidden-tactic note:** no `decide`, no `native_decide` anywhere in the +file; the proofs are `field_simp` + `ring`/`ring_nf` (plus two `rfl`s and +two inductions). The axiom audit below confirms nothing beyond the +standard three. + +## Claims attacked + +1. **005 §(ii)-A Steps 2–4 as ring-model statements**: the pair collapse + (R₁R₂ = Rot_A(2α), R₀R₂ = Rot_B(−2β)), the closed form + u = R₀ ∘ Rot_A(2aα) ∘ Rot_B(−2bβ) = (z ↦ μ·conj z + w), and the eight + formal identities (glide facts μ = δ², Im(conj δ·τ) = 0; I1, I2, I3, + I4; D1, D2), with the (a,b)-dependence carried by formal variables P, Q + in the exponent role — one check for all integers (a,b). Attack = + make Lean's kernel accept them. A wrong sign or a mis-transcribed + closed form would have refused to close under `ring`. +2. **The informal→formal statement mapping** for that block (the + FORMALIZE.md bridge caveat): that the Lean statements are 005's + statements and not a narrowing. Attacked by the encoding table below — + which, per the lane, must be re-checked by a non-formalizer, so it is + flagged **PENDING INDEPENDENT STATEMENT REVIEW**. + +Not attacked, deliberately: **the BRIDGE (geometry = ring model) is +explicitly OUT of scope — 005/008 carry it.** Concretely: that the +circumdiameter-1 normalization and the three base reflections are what +005 Steps 1–2 say they are geometrically, and that `proj`/`m` are the +harness's corridor quantities, remains informal — cross-checked in 005 +(23-member exact specialization + 400-point float check) and 008 (hand +re-derivation + exact off-torus recomposition on 40 members), and stated +verbatim in the module docstring. Also not attacked: Lemmas C/D (L1 was +009), the mod-360 case tree, the necessity theorem, the sufficiency +segments — the analytic layer stays at `VERIFIED`. + +## Refutations found + +None. Every identity of 005's block closed in the kernel with no +hypothesis strengthened and no case dropped. One **generalization** +(a delta in the safe direction, not a narrowing) was found by the +formalization itself: + +- **The block is homogeneous in i.** Every identity holds over an + arbitrary field for ANY nonzero `i`, with no relation `i² = −1`: + both sides of each identity carry the same total 1/(2i)-degree, so `i` + scales out. Discovered because `ring_nf` closed every goal without + ever consuming `i² = −1`; the hypothesis was then removed from the + statements. 005's Q(i)-coefficient ring is the instance + `i := Complex.I` — every 005 use case is covered, and the certificate + is strictly stronger. (Also confirmed numerically at random points + with i = 0.7 − 1.3j before the hypothesis was dropped.) + +A tooling non-finding worth the ledger (queue item 15 flagged this as the +risk): **no Laurent-polynomial or `AddMonoidAlgebra` infrastructure was +needed.** Universally quantifying over a field K and nonzero generators +x, y, P, Q subsumes the Laurent ring (instantiate K at the fraction field +of ℚ[x,y,P,Q] to recover the formal identity; at ℂ on the torus to +recover the geometry), and `field_simp` + `ring_nf` close every identity +in seconds. The anticipated mathlib API fight never happened. + +## Claims that survive + +### Tier reached: A + B + C, all green + +- **Tier A** (closed-form composition): `pair_collapse_A`, + `pair_collapse_B`, `half_word_closed_form` — plus, beyond the asked + scope, `iterate_collapse_A`, `iterate_collapse_B`, `half_word_letters`: + the genuine letter word R₀ (R₁R₂)^a (R₀R₂)^b, as an a- and b-fold + function iterate, equals the closed form with P = x^a, Q = y^b as + actual powers, for every a b : ℕ — this formalizes the power-collapse + step of 005 Steps 1–2 *inside the ring model*, shrinking the informal + bridge to the base-geometry mapping only. +- **Tier B**: `mu_eq_delta_sq`, `tau_parallel_axis`, `mu_mul_conj_mu`, + `half_word_sq_is_translation` (u² = translation by τ), `glide_action` + (p ∘ u = 2m − p, the corridor-reduction identity 008 §1 verified), and + `identity_I1`, `identity_I2`, `identity_I3`, `identity_I4`, + `identity_D1`, `identity_D2`. +- **Tier C** (specialization): `half_word_closed_form_spec`, + `identity_I1_spec` … `identity_I4_spec`, `glide_facts_spec` — P := x^a, + Q := y^b for arbitrary a b : ℤ (zpow). In this encoding 005's + ring-homomorphism specialization is *instantiation of universally + quantified variables*, so each corollary is a one-line application and + the "formal zero ⟹ geometric zero" direction is inherited from the + ∀-statement rather than re-argued. + +### The encoding (formal statement ↔ 005's objects) — **PENDING INDEPENDENT STATEMENT REVIEW** + +All in `formal/BilliardsFormal/LaurentBlock.lean`, namespace +`BilliardsFormal.Laurent`, over `{K : Type*} [Field K]` with hypotheses +`x ≠ 0`, `y ≠ 0`, `P ≠ 0`, `Q ≠ 0`, `i ≠ 0`, `(2 : K) ≠ 0`: + +| 005 / `plaw_general.py` object | formal counterpart | delta | +|---|---|---| +| ring Q(i)[x^±, y^±, P^±, Q^±], x = e^{iα}, y = e^{iβ}, P = e^{iaα}, Q = e^{ibβ} | arbitrary field K, generators x y P Q i ≠ 0, statements ∀-quantified | generalization: any K, any nonzero i (no i² = −1; see above); Laurent ring recovered by instantiation | +| monomial (ex, ey, eP, eQ) | `x^ex * y^ey * P^eP * Q^eQ` | none; exponent lattice = actual exponents | +| sin/cos of ex·α + ey·β + eP·aα + eQ·bβ | `sinF z i = (z - z⁻¹)/(2*i)`, `cosF z = (z + z⁻¹)/2` at the corresponding monomial z | none (Euler form; `sinF` means sine only at i = imaginary unit — bridge note in docstring) | +| conj (exponent negation + coefficient conjugation) | substitution `(x,y,P,Q,i) ↦ (x⁻¹,y⁻¹,P⁻¹,Q⁻¹,−i)` | none on the torus (008 §2's star involution); a point is carried as a pair `(z, conj z)`, conj supplied as the second component / the substituted expression | +| B = sin(α+β), C = sin β·e^{iα}, A₁ = B(1−y⁻²) | `B x y i`, `C x y i`, `A1 x y i` | none (defs transcribe `closed_forms()` monomial by monomial) | +| μ = y⁻²P⁻²Q², δ = y⁻¹P⁻¹Q, w = B(1−y⁻²+y⁻²P⁻²−μ), τ = w + μ·conj w | `mu y P Q`, `delta y P Q`, `w x y P Q i`, `tau x y P Q i` | none | +| p(v) = Im(conj δ·v); m = ½·Im(conj δ·(w−τ/2)) | `proj y P Q i v vc` with vc = conj v; `m x y P Q i` | none (Im ζ = (ζ − conj ζ)/(2i), pair convention) | +| v₂ = B + y⁻²(P⁻²·conj C − B) | `v2 x y P i` | Q dropped from the signature (v₂ does not depend on Q) | +| R₂ = conj; R₁ = e^{2iα}·conj; R₀ = B + e^{−2iβ}·conj(·−B); Rot_A(2aα); Rot_B(−2bβ) | `reflAB`, `reflCA`, `reflBC`, `rotA`, `rotB` on pairs K × K | each map's second component is the conjugate substitution of its first, syntactically | +| specialization P → x^a, Q → y^b by ring homomorphism | instantiation of the ∀-quantified P, Q at `x^(a:ℤ)`, `y^(b:ℤ)` | mechanism differs (instantiation vs ring hom); conclusion identical, and the letters theorems additionally prove the a b : ℕ power-collapse | + +### The formal statements (verbatim, main theorems) + +```lean +theorem half_word_closed_form (x y P Q i : K) + (hx : x ≠ 0) (hy : y ≠ 0) (hP : P ≠ 0) (hQ : Q ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) (p : K × K) : + reflBC x y i (rotA P (rotB x y Q i p)) = + (mu y P Q * p.2 + w x y P Q i, + mu y⁻¹ P⁻¹ Q⁻¹ * p.1 + w x⁻¹ y⁻¹ P⁻¹ Q⁻¹ (-i)) + +theorem half_word_letters (x y i : K) (a b : ℕ) + (hx : x ≠ 0) (hy : y ≠ 0) (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) + (p : K × K) : + reflBC x y i + ((fun q => reflCA x (reflAB q))^[a] + ((fun q => reflBC x y i (reflAB q))^[b] p)) = + (mu y (x ^ a) (y ^ b) * p.2 + w x y (x ^ a) (y ^ b) i, + mu y⁻¹ (x ^ a)⁻¹ (y ^ b)⁻¹ * p.1 + + w x⁻¹ y⁻¹ (x ^ a)⁻¹ (y ^ b)⁻¹ (-i)) + +theorem identity_I1 (x y P Q i : K) (hx : x ≠ 0) (hy : y ≠ 0) + (hP : P ≠ 0) (hQ : Q ≠ 0) (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) : + m x y P Q i - proj y P Q i (C x y i) (C x⁻¹ y⁻¹ (-i)) = + -(cosF P * sinF x i * sinF Q i + + sinF P i * sinF y i * cosF (x * y * Q)) + +theorem identity_I2 ⋯ : + proj y P Q i (A1 x y i) (A1 x⁻¹ y⁻¹ (-i)) - m x y P Q i = + cosF P * sinF (y * Q) i * sinF (x * y) i + +theorem identity_I3 ⋯ : + proj y P Q i (B x y i) (B x⁻¹ y⁻¹ (-i)) - m x y P Q i = + cosF (y * Q) * sinF P i * sinF (x * y) i + +theorem identity_I4 ⋯ : + proj y P Q i (v2 x y P i) (v2 x⁻¹ y⁻¹ P⁻¹ (-i)) - m x y P Q i = + cosF P * sinF x i * sinF Q i - + sinF P i * sinF y i * cosF (x * y * Q) + +theorem identity_D1 ⋯ : + proj y P Q i (B x y i) (B x⁻¹ y⁻¹ (-i)) - + proj y P Q i (C x y i) (C x⁻¹ y⁻¹ (-i)) = + sinF x i * sinF (P * Q⁻¹) i + +theorem identity_D2 ⋯ : + proj y P Q i (A1 x y i) (A1 x⁻¹ y⁻¹ (-i)) - + proj y P Q i (C x y i) (C x⁻¹ y⁻¹ (-i)) = + -(sinF y i * sinF (x⁻¹ * y⁻¹ * P * Q⁻¹) i) + +theorem tau_parallel_axis ⋯ : + proj y P Q i (tau x y P Q i) (tau x⁻¹ y⁻¹ P⁻¹ Q⁻¹ (-i)) = 0 + +theorem mu_eq_delta_sq (y P Q : K) : mu y P Q = delta y P Q ^ 2 + +theorem glide_action ⋯ (z zc : K) : + proj y P Q i (mu y P Q * zc + w x y P Q i) + (mu y⁻¹ P⁻¹ Q⁻¹ * z + w x⁻¹ y⁻¹ P⁻¹ Q⁻¹ (-i)) = + 2 * m x y P Q i - proj y P Q i z zc +``` + +(elided `⋯` = the same six hypotheses as `identity_I1`; full text of all +23 theorems, including the Tier C `*_spec` corollaries at a b : ℤ, in the +source file.) + +### Build and axiom audit + +- `lake build`: `Build completed successfully (8658 jobs)`; + `BilliardsFormal.LaurentBlock` elaborates in ~24 s; zero `sorry`, zero + warnings; the root module imports both `CotMonotone` and `LaurentBlock`. +- `lake env lean AxiomCheck.lean`: all 23 `Laurent` theorems (and 009's + two) depend on exactly `[propext, Classical.choice, Quot.sound]` — no + `sorryAx`, no `Lean.ofReduceBool` (i.e. no `native_decide`). +- Both outputs appended to `formal/lake-build.log` (timestamped block). +- Toolchain pins unchanged: `lean-toolchain` = `leanprover/lean4:v4.32.2`, + mathlib tag `v4.32.2` = rev `905b95818eb32af7874a58b427f50c1711a5e96c` + in `lakefile.toml` / `lake-manifest.json`. + +### What the certificate does and does not cover + +Covered: exactly the formal statements above — 005's identity block and +closed-form composition in the ring model, for all formal P, Q (hence all +integer (a,b) by Tier C), over any field with the six nonzeroness +hypotheses. Not covered (unchanged from 005/008's own scoping, and per +the lane): **the bridge (geometry = ring model) is explicitly OUT of +scope — 005/008 carry it**; the necessity case tree, Lemmas C/D, the +sufficiency segments, and every numerical claim of 005 remain `VERIFIED`, +not `FORMALIZED`. Per `docs/FORMALIZE.md`, this pass is complete only +after (a) an independent statement review of the encoding table above and +(b) an independent rebuild — neither is done by this record. + +## Leads generated + +1. **Complete the FORMALIZED pass** (the 010 pattern): an independent + agent (not this one) (a) re-runs `lake build` + `AxiomCheck.lean` from + the committed source in a fresh container (definite outcome: green or + not), and (b) reviews the encoding table hypothesis-by-hypothesis + against 005 §(ii)-A and `plaw_general.closed_forms()` — the + high-value checks are the conj-as-substitution convention, the pair + second components, and the sinF/cosF monomial dictionary (definite + outcome: sign-off or a filed delta). +2. **The i-homogeneity is a checkable statement about 005's own ring:** + every coefficient in `plaw_general.py`'s eight zero-polynomial checks + should live in ℚ·i^k with k determined by the term's sin-depth, i.e. + the Q(i) coefficients are never genuinely mixed. One afternoon Python + check of the coefficient dictionaries; if true, 005's ring can be + restated over ℚ with i a formal unit — worth knowing before any + future port (definite outcome: true/false). +3. **Formalize Lemma D and the L1 → C/D chain** (009 lead 2, still + open): with L1 and now the block formal, Lemmas C/D are the remaining + informal layer between the certificates and 005's necessity theorem. +4. **Shrink the bridge with a torus-evaluation lemma:** over ℂ, + `sinF (Complex.exp (θ * I)) I = Complex.sin θ` (Euler) would connect + `sinF`/`cosF` to actual trigonometry, moving the "sinF means sine" + line of the mapping table from prose into the kernel. Small, definite. +5. **Port the skeleton to the design-track families** (005 lead 5): the + file contains nothing billiards-specific except the three base + reflections; any family (w₀B₁^aB₂^b)² with vertex-pivoting blocks + gets the same `half_word_letters`-style theorem by editing the three + `refl*` definitions (definite outcome per family: builds or surfaces + a structural mismatch). + +## References + +- `problems/billiards-triangles/attempts/005-complete-death-law-theorem.md` + §(ii)-A (the block formalized), §Approach (the formal-ring method). +- `problems/billiards-triangles/attempts/008-skeptic-review-of-005.md` + §§1–2 (skeptic confirmation; the off-torus pair algebra and the + specialization-soundness argument this encoding leans on). +- `problems/billiards-triangles/attempts/009-lean-pilot-cot-lemma.md` and + `010-statement-review-of-009.md` (lane precedent; toolchain; the + AxiomCheck nit fixed here). +- `problems/billiards-triangles/explore/plaw_general.py` (`closed_forms`, + `cmd_formal` — the proof object transcribed). +- `docs/FORMALIZE.md` (the lane: qualification, bridge caveat, three-part + pass). +- mathlib4, tag v4.32.2, github.com/leanprover-community/mathlib4 + (rev `905b95818eb32af7874a58b427f50c1711a5e96c`). No papers consulted; + no machine transcriptions. diff --git a/problems/billiards-triangles/formal/AxiomCheck.lean b/problems/billiards-triangles/formal/AxiomCheck.lean new file mode 100644 index 0000000..daa3ab6 --- /dev/null +++ b/problems/billiards-triangles/formal/AxiomCheck.lean @@ -0,0 +1,40 @@ +/- +Axiom audit for the BilliardsFormal certificates. Run with +`lake env lean AxiomCheck.lean` from `problems/billiards-triangles/formal/`; +every theorem must report exactly [propext, Classical.choice, Quot.sound] +(and never `sorryAx`). Not part of the library build. +-/ +import BilliardsFormal + +-- 009: Lemma L1 +#print axioms BilliardsFormal.cot_lemma_L1 +#print axioms BilliardsFormal.strictAntiOn_id_mul_cot + +-- 011, Tier A +#print axioms BilliardsFormal.Laurent.pair_collapse_A +#print axioms BilliardsFormal.Laurent.pair_collapse_B +#print axioms BilliardsFormal.Laurent.half_word_closed_form +#print axioms BilliardsFormal.Laurent.iterate_collapse_A +#print axioms BilliardsFormal.Laurent.iterate_collapse_B +#print axioms BilliardsFormal.Laurent.half_word_letters + +-- 011, Tier B +#print axioms BilliardsFormal.Laurent.mu_eq_delta_sq +#print axioms BilliardsFormal.Laurent.tau_parallel_axis +#print axioms BilliardsFormal.Laurent.mu_mul_conj_mu +#print axioms BilliardsFormal.Laurent.half_word_sq_is_translation +#print axioms BilliardsFormal.Laurent.glide_action +#print axioms BilliardsFormal.Laurent.identity_I1 +#print axioms BilliardsFormal.Laurent.identity_I2 +#print axioms BilliardsFormal.Laurent.identity_I3 +#print axioms BilliardsFormal.Laurent.identity_I4 +#print axioms BilliardsFormal.Laurent.identity_D1 +#print axioms BilliardsFormal.Laurent.identity_D2 + +-- 011, Tier C +#print axioms BilliardsFormal.Laurent.half_word_closed_form_spec +#print axioms BilliardsFormal.Laurent.identity_I1_spec +#print axioms BilliardsFormal.Laurent.identity_I2_spec +#print axioms BilliardsFormal.Laurent.identity_I3_spec +#print axioms BilliardsFormal.Laurent.identity_I4_spec +#print axioms BilliardsFormal.Laurent.glide_facts_spec diff --git a/problems/billiards-triangles/formal/BilliardsFormal.lean b/problems/billiards-triangles/formal/BilliardsFormal.lean index cf2f6f4..0b6bdf2 100644 --- a/problems/billiards-triangles/formal/BilliardsFormal.lean +++ b/problems/billiards-triangles/formal/BilliardsFormal.lean @@ -1 +1,2 @@ import BilliardsFormal.CotMonotone +import BilliardsFormal.LaurentBlock diff --git a/problems/billiards-triangles/formal/BilliardsFormal/LaurentBlock.lean b/problems/billiards-triangles/formal/BilliardsFormal/LaurentBlock.lean new file mode 100644 index 0000000..f99e5e3 --- /dev/null +++ b/problems/billiards-triangles/formal/BilliardsFormal/LaurentBlock.lean @@ -0,0 +1,423 @@ +/- +The Laurent-identity block of attempt record 005 +(`problems/billiards-triangles/attempts/005-complete-death-law-theorem.md`, +sections (ii)-A Steps 2-4), reviewed and confirmed in 008 (sections 1-2): +the closed-form composition of the half word of W(a,b) and the identities +I1, I2, I3, I4, D1, D2 plus the glide facts, with the family parameters +(a, b) carried by FORMAL variables, so that one kernel-checked proof covers +every integer pair (a, b) at once (005's formal-parameter specialization). + +## The encoding (the formal statement IS this model — read before trusting) + +005 works in the Laurent ring Q(i)[x^±, y^±, P^±, Q^±] where, on the torus, +x = e^{iα}, y = e^{iβ}, P = e^{iaα}, Q = e^{ibβ}. Here that ring is +represented by an arbitrary field `K` with distinguished elements +`x y P Q i : K` subject only to `x ≠ 0`, `y ≠ 0`, `P ≠ 0`, `Q ≠ 0`, +`i ≠ 0`, `(2 : K) ≠ 0`. Note `i` needs no relation `i² = -1`: every +identity in the block turned out to be homogeneous in `i` (each side has +the same total `1/(2i)`-degree), so it holds for ANY nonzero `i` — a +strict generalization of 005's Q(i) setting, discovered because `ring` +closed every goal without ever consuming `i² = -1`. Every theorem is +universally quantified over `K` and these generators, so instantiating +`K := FractionField` of the Laurent ring recovers the formal identity, and +instantiating `K := ℂ`, `x := exp (I*α)`, `y := exp (I*β)`, +`P := exp (I*a*α)`, `Q := exp (I*b*β)`, `i := Complex.I` recovers the +geometric specialization. Consequently: + +* a monomial `x^ex * y^ey * P^eP * Q^eQ` stands for the unit + `e^{i(ex·α + ey·β + eP·aα + eQ·bβ)}`; the (a,b)-dependence lives in the + formal generators `P`, `Q` only (005's exponent-lattice move); +* formal sine and cosine of the angle whose unit is `z` are + `sinF z i = (z - z⁻¹) / (2*i)` and `cosF z = (z + z⁻¹) / 2`; +* complex conjugation is the substitution + `(x, y, P, Q, i) ↦ (x⁻¹, y⁻¹, P⁻¹, Q⁻¹, -i)` — on the torus all four + generators are units of modulus 1, so inversion is conjugation + (008 §2's star involution). A geometric point is carried as a PAIR + `(z, conj z) : K × K`, and each map on pairs has as second component the + conjugate substitution of its first (an invariant that is syntactically + visible in the definitions below, and proved for the composite maps); +* `Im ζ = (ζ - conj ζ) / (2*i)`, so the glide-axis projection is + `proj … v vc = (conj δ * v - δ * vc) / (2*i)` with `vc = conj v`. + +The definitions `B, C, A1, mu, delta, w, tau, m, v2` below transcribe +`closed_forms()` of `explore/plaw_general.py` (005's proof object, +independently re-derived in 008) monomial by monomial. + +## What is proved (tiers as in the queue-15 task) + +* Tier A — `pair_collapse_A`, `pair_collapse_B`, `half_word_closed_form`: + R₁R₂ = Rot_A(2α), R₀R₂ = Rot_B(−2β) (005 Step 2), and + R₀ ∘ Rot_A ∘ Rot_B = (z ↦ μ·conj z + w) with μ, w the closed forms + (005 Step 3). `iterate_collapse_A/B` and `half_word_letters` extend this + to the genuine letter word R₀ (R₁R₂)^a (R₀R₂)^b for every a b : ℕ, with + P = x^a, Q = y^b appearing as actual powers. +* Tier B — `mu_eq_delta_sq`, `tau_parallel_axis`, `mu_mul_conj_mu`, + `half_word_sq_is_translation`, `glide_action`, and `identity_I1 … I4`, + `identity_D1`, `identity_D2`: 005's Step 4 identity list, verbatim. +* Tier C — the `*_spec` corollaries: specialization to arbitrary integer + exponents `P := x^a`, `Q := y^b` (`a b : ℤ`). In this encoding 005's + ring-homomorphism specialization is instantiation of universally + quantified variables, so each corollary is a one-line application. + +## What is NOT proved (the bridge) + +The bridge (geometry = ring model) is explicitly OUT of scope — 005/008 +carry it. Concretely, the following remain informal, cross-checked in +005 (23-member exact specialization, 400-point float check) and 008 +(hand re-derivation + exact off-torus recomposition, 40 members): +the circumdiameter-1 normalization A = 0, B = sin(α+β), C = sin β·e^{iα}; +that `reflAB`, `reflCA`, `reflBC` are the reflections in the sides AB, CA, +BC of that triangle; and that the unfolding corridor quantities of the +harness are the `proj`/`m` data below. A compiling proof here says nothing +about billiards until that mapping is reviewed against 005 §Step 1-3. + +Zero `sorry`; no axioms beyond mathlib (audit in `lake-build.log`). +Toolchain pinned in `lean-toolchain` (leanprover/lean4:v4.32.2), mathlib +pinned in `lakefile.toml` / `lake-manifest.json` (tag v4.32.2). +-/ +import Mathlib + +namespace BilliardsFormal +namespace Laurent + +variable {K : Type*} [Field K] + +/-! ### The formal trigonometric dictionary -/ + +/-- Formal sine of the angle whose unit is `z`: on the torus +`z = e^{iθ}` this is `(e^{iθ} - e^{-iθ})/(2i) = sin θ`. -/ +def sinF (z i : K) : K := (z - z⁻¹) / (2 * i) + +/-- Formal cosine of the angle whose unit is `z`. -/ +def cosF (z : K) : K := (z + z⁻¹) / 2 + +/-! ### Vertices and glide data (005 Step 3 / `plaw_general.closed_forms`) -/ + +/-- Vertex `B = sin(α+β)` (circumdiameter-1 normalization; real). -/ +def B (x y i : K) : K := sinF (x * y) i + +/-- Vertex `C = sin β · e^{iα}`. -/ +def C (x y i : K) : K := sinF y i * x + +/-- `A₁ = R₀(A) = B · (1 - y⁻²)`, the A-fan pivot. -/ +def A1 (x y i : K) : K := B x y i * (1 - y⁻¹ ^ 2) + +/-- Antilinear multiplier of the half word: `μ = e^{-2iaα + 2i(b-1)β} += y⁻² P⁻² Q²`. -/ +def mu (y P Q : K) : K := y⁻¹ ^ 2 * P⁻¹ ^ 2 * Q ^ 2 + +/-- Glide-axis direction `δ = e^{i(-aα + (b-1)β)} = y⁻¹ P⁻¹ Q`; +`δ² = μ`. -/ +def delta (y P Q : K) : K := y⁻¹ * P⁻¹ * Q + +/-- Translation part of the half word: +`w = B · (1 - y⁻² + y⁻²P⁻² - μ)`. -/ +def w (x y P Q i : K) : K := + B x y i * (1 - y⁻¹ ^ 2 + y⁻¹ ^ 2 * P⁻¹ ^ 2 - mu y P Q) + +/-- Glide translation `τ = w + μ · conj w` (conjugation = substitution). -/ +def tau (x y P Q i : K) : K := + w x y P Q i + mu y P Q * w x⁻¹ y⁻¹ P⁻¹ Q⁻¹ (-i) + +/-- Projection onto the normal of the glide axis: +`p(v) = Im(conj δ · v) = (conj δ · v - δ · conj v)/(2i)`, with the +conjugate `vc = conj v` supplied explicitly (pair convention). -/ +def proj (y P Q i v vc : K) : K := + (delta y⁻¹ P⁻¹ Q⁻¹ * v - delta y P Q * vc) / (2 * i) + +/-- Axis offset `m = ½ · Im(conj δ · (w - τ/2))`. -/ +def m (x y P Q i : K) : K := + proj y P Q i (w x y P Q i - tau x y P Q i / 2) + (w x⁻¹ y⁻¹ P⁻¹ Q⁻¹ (-i) - tau x⁻¹ y⁻¹ P⁻¹ Q⁻¹ (-i) / 2) / 2 + +/-- The C-image endpoint of gate `2a+2`: +`v₂ = R₀(Rot_A(2aα) C) = B + y⁻²(P⁻² · conj C - B)` (005's I4 anchor). -/ +def v2 (x y P i : K) : K := + B x y i + y⁻¹ ^ 2 * (P⁻¹ ^ 2 * C x⁻¹ y⁻¹ (-i) - B x y i) + +/-! ### The maps, on pairs `(z, conj z)` + +Each second component is the conjugate substitution of the first — +generators inverted, `i` negated, components swapped (for the +orientation-reversing reflections) or kept (for the rotations). -/ + +/-- `R₂`: reflection in side AB (the real axis): `z ↦ conj z`. -/ +def reflAB (p : K × K) : K × K := (p.2, p.1) + +/-- `R₁`: reflection in side CA: `z ↦ x² · conj z` (`x² = e^{2iα}`). -/ +def reflCA (x : K) (p : K × K) : K × K := (x ^ 2 * p.2, x⁻¹ ^ 2 * p.1) + +/-- `R₀`: reflection in side BC: `z ↦ B + y⁻² (conj z - B)` +(`y⁻² = e^{-2iβ}`, and `conj B = B`). -/ +def reflBC (x y i : K) (p : K × K) : K × K := + (B x y i + y⁻¹ ^ 2 * (p.2 - B x y i), + B x⁻¹ y⁻¹ (-i) + (y⁻¹)⁻¹ ^ 2 * (p.1 - B x⁻¹ y⁻¹ (-i))) + +/-- `Rot_A(2aα)`: rotation about `A = 0` by `2aα`: `z ↦ P² z`. -/ +def rotA (P : K) (p : K × K) : K × K := (P ^ 2 * p.1, P⁻¹ ^ 2 * p.2) + +/-- `Rot_B(-2bβ)`: rotation about `B` by `-2bβ`: `z ↦ B + Q⁻²(z - B)`. -/ +def rotB (x y Q i : K) (p : K × K) : K × K := + (B x y i + Q⁻¹ ^ 2 * (p.1 - B x y i), + B x⁻¹ y⁻¹ (-i) + (Q⁻¹)⁻¹ ^ 2 * (p.2 - B x⁻¹ y⁻¹ (-i))) + +/-! ### Tier A: pair collapse and the closed-form composition (005 Steps 2-3) -/ + +/-- 005 Step 2, first half: `R₁ ∘ R₂ = Rot_A(2α)` (definitional). -/ +theorem pair_collapse_A (x : K) (p : K × K) : + reflCA x (reflAB p) = rotA x p := rfl + +/-- 005 Step 2, second half: `R₀ ∘ R₂ = Rot_B(-2β)` (definitional: +`rotB` at `Q := y`). -/ +theorem pair_collapse_B (x y i : K) (p : K × K) : + reflBC x y i (reflAB p) = rotB x y y i p := rfl + +/-- 005 Step 3, the closed form of the half word: as maps on pairs, +`R₀ ∘ Rot_A(2aα) ∘ Rot_B(-2bβ) = (z ↦ μ · conj z + w)`, with `μ`, `w` +the closed forms and the parameters entering only through `P`, `Q`. -/ +theorem half_word_closed_form (x y P Q i : K) + (hx : x ≠ 0) (hy : y ≠ 0) (hP : P ≠ 0) (hQ : Q ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) (p : K × K) : + reflBC x y i (rotA P (rotB x y Q i p)) = + (mu y P Q * p.2 + w x y P Q i, + mu y⁻¹ P⁻¹ Q⁻¹ * p.1 + w x⁻¹ y⁻¹ P⁻¹ Q⁻¹ (-i)) := by + rw [Prod.ext_iff] + constructor <;> + · simp only [reflBC, rotA, rotB, mu, w, B, sinF, inv_inv] + field_simp + ring_nf + +/-- The letter word `(R₁R₂)^a` collapses to `Rot_A(2aα)` with `P = x^a` +as an actual power (the power step of 005 Step 1/2, formal). -/ +theorem iterate_collapse_A (x : K) (a : ℕ) (p : K × K) : + (fun q => reflCA x (reflAB q))^[a] p = rotA (x ^ a) p := by + induction a generalizing p with + | zero => simp [rotA] + | succ n ih => + rw [Function.iterate_succ_apply', ih, pair_collapse_A] + rw [Prod.ext_iff] + constructor <;> + · simp only [rotA, mul_inv, pow_succ] + ring + +/-- The letter word `(R₀R₂)^b` collapses to `Rot_B(-2bβ)` with `Q = y^b` +as an actual power. -/ +theorem iterate_collapse_B (x y i : K) (b : ℕ) (p : K × K) : + (fun q => reflBC x y i (reflAB q))^[b] p = rotB x y (y ^ b) i p := by + induction b generalizing p with + | zero => simp [rotB] + | succ n ih => + rw [Function.iterate_succ_apply', ih, pair_collapse_B] + rw [Prod.ext_iff] + constructor <;> + · simp only [rotB, mul_inv, pow_succ, inv_inv] + ring + +/-- The full letter half word `R₀ (R₁R₂)^a (R₀R₂)^b` in closed form, for +every `a b : ℕ`: `u(z) = μ conj z + w` with `P = x^a`, `Q = y^b`. -/ +theorem half_word_letters (x y i : K) (a b : ℕ) + (hx : x ≠ 0) (hy : y ≠ 0) (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) + (p : K × K) : + reflBC x y i + ((fun q => reflCA x (reflAB q))^[a] + ((fun q => reflBC x y i (reflAB q))^[b] p)) = + (mu y (x ^ a) (y ^ b) * p.2 + w x y (x ^ a) (y ^ b) i, + mu y⁻¹ (x ^ a)⁻¹ (y ^ b)⁻¹ * p.1 + + w x⁻¹ y⁻¹ (x ^ a)⁻¹ (y ^ b)⁻¹ (-i)) := by + rw [iterate_collapse_B, iterate_collapse_A, + half_word_closed_form x y (x ^ a) (y ^ b) i hx hy + (pow_ne_zero a hx) (pow_ne_zero b hy) hI h2 p] + +/-! ### Tier B: the glide facts and I1-I4, D1, D2 (005 Step 4) -/ + +/-- Glide fact: `μ = δ²`. -/ +theorem mu_eq_delta_sq (y P Q : K) : mu y P Q = delta y P Q ^ 2 := by + simp only [mu, delta]; ring + +/-- Glide fact: `τ` is parallel to the glide axis, `Im(conj δ · τ) = 0`. -/ +theorem tau_parallel_axis (x y P Q i : K) + (hx : x ≠ 0) (hy : y ≠ 0) (hP : P ≠ 0) (hQ : Q ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) : + proj y P Q i (tau x y P Q i) (tau x⁻¹ y⁻¹ P⁻¹ Q⁻¹ (-i)) = 0 := by + simp only [proj, tau, mu, w, delta, B, sinF, inv_inv, neg_neg] + field_simp + ring_nf + +/-- `μ` is a unit with `μ · conj μ = 1` (the half word is an isometry). -/ +theorem mu_mul_conj_mu (y P Q : K) + (hy : y ≠ 0) (hP : P ≠ 0) (hQ : Q ≠ 0) : + mu y P Q * mu y⁻¹ P⁻¹ Q⁻¹ = 1 := by + simp only [mu, inv_inv] + field_simp + +/-- The square of the half word is the translation by `τ`: +`u(u(z)) = z + τ` (005 Step 3's "u² is automatically a translation"). -/ +theorem half_word_sq_is_translation (x y P Q i : K) + (hx : x ≠ 0) (hy : y ≠ 0) (hP : P ≠ 0) (hQ : Q ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) (z : K) : + mu y P Q * (mu y⁻¹ P⁻¹ Q⁻¹ * z + w x⁻¹ y⁻¹ P⁻¹ Q⁻¹ (-i)) + + w x y P Q i = z + tau x y P Q i := by + simp only [tau, mu, w, B, sinF, inv_inv] + field_simp + ring + +/-- The glide action on the axis-normal projection: `p ∘ u = 2m - p` +(the identity behind 005's corridor reduction, 008 §1's `p(u(z)) + p(z) +- 2m = 0`). -/ +theorem glide_action (x y P Q i : K) + (hx : x ≠ 0) (hy : y ≠ 0) (hP : P ≠ 0) (hQ : Q ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) (z zc : K) : + proj y P Q i (mu y P Q * zc + w x y P Q i) + (mu y⁻¹ P⁻¹ Q⁻¹ * z + w x⁻¹ y⁻¹ P⁻¹ Q⁻¹ (-i)) = + 2 * m x y P Q i - proj y P Q i z zc := by + simp only [proj, m, tau, mu, w, delta, B, sinF, inv_inv, neg_neg] + field_simp + ring_nf + +/-- **I1** (005 (ii)-A Step 4): `m - p(C) = -[cos(aα) sin α sin(bβ) ++ sin(aα) sin β cos(α + (b+1)β)]`. -/ +theorem identity_I1 (x y P Q i : K) + (hx : x ≠ 0) (hy : y ≠ 0) (hP : P ≠ 0) (hQ : Q ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) : + m x y P Q i - proj y P Q i (C x y i) (C x⁻¹ y⁻¹ (-i)) = + -(cosF P * sinF x i * sinF Q i + + sinF P i * sinF y i * cosF (x * y * Q)) := by + simp only [proj, m, tau, mu, w, delta, B, C, sinF, cosF, inv_inv, + neg_neg, mul_inv] + field_simp + ring_nf + +/-- **I2**: `p(A₁) - m = cos(aα) sin((b+1)β) sin(α+β)`. -/ +theorem identity_I2 (x y P Q i : K) + (hx : x ≠ 0) (hy : y ≠ 0) (hP : P ≠ 0) (hQ : Q ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) : + proj y P Q i (A1 x y i) (A1 x⁻¹ y⁻¹ (-i)) - m x y P Q i = + cosF P * sinF (y * Q) i * sinF (x * y) i := by + simp only [proj, m, tau, mu, w, delta, A1, B, sinF, cosF, inv_inv, + neg_neg, mul_inv] + field_simp + ring_nf + +/-- **I3**: `p(B) - m = cos((b+1)β) sin(aα) sin(α+β)`. -/ +theorem identity_I3 (x y P Q i : K) + (hx : x ≠ 0) (hy : y ≠ 0) (hP : P ≠ 0) (hQ : Q ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) : + proj y P Q i (B x y i) (B x⁻¹ y⁻¹ (-i)) - m x y P Q i = + cosF (y * Q) * sinF P i * sinF (x * y) i := by + simp only [proj, m, tau, mu, w, delta, B, sinF, cosF, inv_inv, + neg_neg, mul_inv] + field_simp + ring_nf + +/-- **I4** (new in 005; the gate-(2a+2) C-image endpoint): +`p(v₂) - m = cos(aα) sin α sin(bβ) - sin(aα) sin β cos(α + (b+1)β)`. +Note I1 `= -(S+T)` and I4 `= S-T` for the same two products. -/ +theorem identity_I4 (x y P Q i : K) + (hx : x ≠ 0) (hy : y ≠ 0) (hP : P ≠ 0) (hQ : Q ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) : + proj y P Q i (v2 x y P i) (v2 x⁻¹ y⁻¹ P⁻¹ (-i)) - m x y P Q i = + cosF P * sinF x i * sinF Q i - + sinF P i * sinF y i * cosF (x * y * Q) := by + simp only [proj, m, tau, mu, w, delta, v2, B, C, sinF, cosF, inv_inv, + neg_neg, mul_inv] + field_simp + ring_nf + +/-- **D1** (collapsing-gate locus): `p(B) - p(C) = sin α sin(aα - bβ)`. -/ +theorem identity_D1 (x y P Q i : K) + (hx : x ≠ 0) (hy : y ≠ 0) (hP : P ≠ 0) (hQ : Q ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) : + proj y P Q i (B x y i) (B x⁻¹ y⁻¹ (-i)) - + proj y P Q i (C x y i) (C x⁻¹ y⁻¹ (-i)) = + sinF x i * sinF (P * Q⁻¹) i := by + simp only [proj, delta, B, C, sinF, inv_inv, mul_inv] + field_simp + ring_nf + +/-- **D2** (collapsing-gate locus): +`p(A₁) - p(C) = -sin β sin((a-1)α - (b+1)β)`. -/ +theorem identity_D2 (x y P Q i : K) + (hx : x ≠ 0) (hy : y ≠ 0) (hP : P ≠ 0) (hQ : Q ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) : + proj y P Q i (A1 x y i) (A1 x⁻¹ y⁻¹ (-i)) - + proj y P Q i (C x y i) (C x⁻¹ y⁻¹ (-i)) = + -(sinF y i * sinF (x⁻¹ * y⁻¹ * P * Q⁻¹) i) := by + simp only [proj, delta, A1, B, C, sinF, inv_inv, mul_inv] + field_simp + ring_nf + +/-! ### Tier C: specialization to integer exponents (005's ring +homomorphism `P → x^a`, `Q → y^b`) + +In this encoding the specialization is instantiation of the universally +quantified generators, so each corollary is a one-line application; the +`a b : ℤ` genericity is inherited, never re-proved. (`zpow_ne_zero`: +in a field `x ≠ 0 → x^(a:ℤ) ≠ 0`.) -/ + +section Specialize + +variable (x y i : K) (a b : ℤ) + +/-- The closed-form composition at `P = x^a`, `Q = y^b`, `a b : ℤ`. -/ +theorem half_word_closed_form_spec (hx : x ≠ 0) (hy : y ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) (p : K × K) : + reflBC x y i (rotA (x ^ a) (rotB x y (y ^ b) i p)) = + (mu y (x ^ a) (y ^ b) * p.2 + w x y (x ^ a) (y ^ b) i, + mu y⁻¹ (x ^ a)⁻¹ (y ^ b)⁻¹ * p.1 + + w x⁻¹ y⁻¹ (x ^ a)⁻¹ (y ^ b)⁻¹ (-i)) := + half_word_closed_form x y (x ^ a) (y ^ b) i hx hy + (zpow_ne_zero a hx) (zpow_ne_zero b hy) hI h2 p + +/-- I1 for every pair of integer exponents. -/ +theorem identity_I1_spec (hx : x ≠ 0) (hy : y ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) : + m x y (x ^ a) (y ^ b) i - + proj y (x ^ a) (y ^ b) i (C x y i) (C x⁻¹ y⁻¹ (-i)) = + -(cosF (x ^ a) * sinF x i * sinF (y ^ b) i + + sinF (x ^ a) i * sinF y i * cosF (x * y * y ^ b)) := + identity_I1 x y (x ^ a) (y ^ b) i hx hy + (zpow_ne_zero a hx) (zpow_ne_zero b hy) hI h2 + +/-- I2 for every pair of integer exponents. -/ +theorem identity_I2_spec (hx : x ≠ 0) (hy : y ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) : + proj y (x ^ a) (y ^ b) i (A1 x y i) (A1 x⁻¹ y⁻¹ (-i)) - + m x y (x ^ a) (y ^ b) i = + cosF (x ^ a) * sinF (y * y ^ b) i * sinF (x * y) i := + identity_I2 x y (x ^ a) (y ^ b) i hx hy + (zpow_ne_zero a hx) (zpow_ne_zero b hy) hI h2 + +/-- I3 for every pair of integer exponents. -/ +theorem identity_I3_spec (hx : x ≠ 0) (hy : y ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) : + proj y (x ^ a) (y ^ b) i (B x y i) (B x⁻¹ y⁻¹ (-i)) - + m x y (x ^ a) (y ^ b) i = + cosF (y * y ^ b) * sinF (x ^ a) i * sinF (x * y) i := + identity_I3 x y (x ^ a) (y ^ b) i hx hy + (zpow_ne_zero a hx) (zpow_ne_zero b hy) hI h2 + +/-- I4 for every pair of integer exponents. -/ +theorem identity_I4_spec (hx : x ≠ 0) (hy : y ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) : + proj y (x ^ a) (y ^ b) i (v2 x y (x ^ a) i) + (v2 x⁻¹ y⁻¹ (x ^ a)⁻¹ (-i)) - + m x y (x ^ a) (y ^ b) i = + cosF (x ^ a) * sinF x i * sinF (y ^ b) i - + sinF (x ^ a) i * sinF y i * cosF (x * y * y ^ b) := + identity_I4 x y (x ^ a) (y ^ b) i hx hy + (zpow_ne_zero a hx) (zpow_ne_zero b hy) hI h2 + +/-- The glide facts for every pair of integer exponents. -/ +theorem glide_facts_spec (hx : x ≠ 0) (hy : y ≠ 0) + (hI : i ≠ 0) (h2 : (2 : K) ≠ 0) : + mu y (x ^ a) (y ^ b) = delta y (x ^ a) (y ^ b) ^ 2 ∧ + proj y (x ^ a) (y ^ b) i (tau x y (x ^ a) (y ^ b) i) + (tau x⁻¹ y⁻¹ (x ^ a)⁻¹ (y ^ b)⁻¹ (-i)) = 0 := + ⟨mu_eq_delta_sq y (x ^ a) (y ^ b), + tau_parallel_axis x y (x ^ a) (y ^ b) i hx hy + (zpow_ne_zero a hx) (zpow_ne_zero b hy) hI h2⟩ + +end Specialize + +end Laurent +end BilliardsFormal diff --git a/problems/billiards-triangles/formal/lake-build.log b/problems/billiards-triangles/formal/lake-build.log index f8d1e11..54ba00a 100644 --- a/problems/billiards-triangles/formal/lake-build.log +++ b/problems/billiards-triangles/formal/lake-build.log @@ -8,3 +8,36 @@ Build completed successfully (8657 jobs). $ lake env lean AxiomCheck.lean # axiom audit of the two theorems 'BilliardsFormal.cot_lemma_L1' depends on axioms: [propext, Classical.choice, Quot.sound] 'BilliardsFormal.strictAntiOn_id_mul_cot' depends on axioms: [propext, Classical.choice, Quot.sound] + +==================================================================== +$ lake build # 2026-08-04T10:13:00Z (after adding BilliardsFormal/LaurentBlock.lean, attempt 011) +$ lean --version: Lean (version 4.32.2, x86_64-unknown-linux-gnu, commit f3b06c705e6c85f5314019d5d3baab0fec5b580c, Release) +$ mathlib rev: 905b95818eb32af7874a58b427f50c1711a5e96c (tag v4.32.2) +Build completed successfully (8658 jobs). + +$ lake env lean AxiomCheck.lean # axiom audit, all BilliardsFormal theorems (009 + 011) +'BilliardsFormal.cot_lemma_L1' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.strictAntiOn_id_mul_cot' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.pair_collapse_A' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.pair_collapse_B' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.half_word_closed_form' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.iterate_collapse_A' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.iterate_collapse_B' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.half_word_letters' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.mu_eq_delta_sq' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.tau_parallel_axis' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.mu_mul_conj_mu' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.half_word_sq_is_translation' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.glide_action' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.identity_I1' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.identity_I2' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.identity_I3' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.identity_I4' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.identity_D1' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.identity_D2' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.half_word_closed_form_spec' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.identity_I1_spec' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.identity_I2_spec' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.identity_I3_spec' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.identity_I4_spec' depends on axioms: [propext, Classical.choice, Quot.sound] +'BilliardsFormal.Laurent.glide_facts_spec' depends on axioms: [propext, Classical.choice, Quot.sound] diff --git a/problems/billiards-triangles/prior-art.json b/problems/billiards-triangles/prior-art.json index ec3431b..9c97a8a 100644 --- a/problems/billiards-triangles/prior-art.json +++ b/problems/billiards-triangles/prior-art.json @@ -1,293 +1,320 @@ { - "problem": "billiards-triangles", - "route_status": "EVIDENCE", - "route_summary": "Word-census route mapped blind to length 26: 376 exact certificates (family orbits to length 66, gamma <= 160 deg), coverage staircase stalling at 135 deg, and a measured hyperbolic (not super-exponential) growth of needed word length vs angle - but certified box width collapses exponentially in length, so length buys angle reach, not area. Negatives are sample-bounded; skeptic-confirmed with one numeric correction.", - "attempts": [ - { - "id": "001", - "file": "attempts/001-word-census-coverage-map.md", - "date": "2026-07-30", - "mode": "blind", - "status": "EVIDENCE", - "mechanism": [ - "unfolding-word-census", - "certified-box-coverage", - "exhaustive-enumeration" - ], - "one_line": "Canonical word census to length 26 with 376 exact certificates: coverage staircase stalls at 135 deg, blind rediscovery of the ~112.5 deg frontier, and a hyperbolic certificate-length growth law along the W(a,b) family (checked to length 66).", - "leak_terms": [ - "112.4989", - "W(a,b)", - "coverage staircase", - "length buys angle reach", - "17,527" - ], - "range": "canonical translation words of even length <= 26 (17,527 words); 1/64-cell midpoint screen of the obtuse half-disk; exact dyadic-box anchors; family rows to W(8,8), gamma <= 160 deg", - "gaps": [ - "sample-bounded-negatives", - "death-angle-laws-unproven" - ] - }, - { - "id": "002", - "file": "attempts/002-skeptic-review-of-001.md", - "date": "2026-07-30", - "mode": "blind", - "status": "VERIFIED_WITH_CORRECTIONS", - "mechanism": [ - "adversarial-verification", - "independent-reimplementation", - "unfolding-word-census" - ], - "one_line": "From-scratch re-implementation of every layer confirms 001 (36/36 sampled certificates at 180 exact apexes, 17,527-word universe reproduced word-for-word, death law survives two out-of-sample prediction rows); certified-area figure corrected to ~4.3e-6.", - "leak_terms": [ - "147.5974", - "157.4976", - "4.3e-6" - ], - "verifies": "001", - "corrections": [ - "certified-area total over all 376 anchors is ~4.3e-6, not ~2.2e-6 (which is the map-anchor subtotal)", - "the 64.3% uncovered figure is of the resolved interior, not the whole obtuse region", - "anchor simulation re-derivation is at the anchor centre; the full 9-sample box attack ran on every 5th/7th record", - "kill-condition verdict is measured only to gamma ~ 160 deg; beyond that the hyperbolic form is the SPECULATION law's extrapolation" - ] - }, - { - "id": "003", - "file": "attempts/003-death-angle-laws.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED", - "mechanism": [ - "symbolic-trig-closed-form", - "certified-interval-inequality", - "out-of-sample-prediction", - "unfolding-word-census" - ], - "one_line": "The W(a,b) death-angle laws of 001 survive both out-of-sample tests (W(10,9) -> 162.90, W(12,12) -> 2160/13, to ~1e-12), and are then derived: exact division-free unfolding gives closed-form binding functions, a general law gamma_d = 180 - 90(a+b)/(a(b+1)) (confirmed out-of-sample on W(5,3) -> 144 and W(4,2) -> 135), and a machine-certified necessity theorem: the corridor is empty at every gamma >= gamma_d, for 15 verified members.", - "leak_terms": [ - "gamma_d = 180 - 90(a+b)/(a(b+1))", - "90(a+b)/(a(b+1))", - "glide reflection reduction", - "W(5,3)", - "144.0000000000", - "deathlaw_", - "Lemma C", - "division-free unfolding" - ], - "range": "necessity theorem machine-certified for the 15 members (2,1),(2,2),(3,2),(3,3),(4,2),(4,3),(4,4),(5,3),(5,4),(5,5),(6,6),(7,7),(8,8),(10,9),(12,12); exact alive certificates at gamma_d - 1e-6 (1e-8 for (3,3),(4,2)) for 8 members; float deaths measured to ~1e-11 for 11 members; parametric statement for all a >= b >= 1, a >= 2, a <= 2b+3 rests on two per-member-verified inputs (identities I1-I3 and Lemma C) labelled SPECULATION in general", - "gaps": [ - "sufficiency-limit-unclosed (death = gamma_d exactly needs alive points approaching gamma_d; certified alive stops at gamma_d - 1e-6)", - "parametric-identities-unproven-in-general (I1-I3 and Lemma C verified per member, 15 members; general (a,b) statement labelled SPECULATION)", - "a-greater-than-2b+3-uncovered (branch analysis breaks; W(6,3)-type members unmeasured)" - ], - "refuted_by": null - }, - { - "id": "004", - "file": "attempts/004-skeptic-review-of-003.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED_WITH_CORRECTIONS", - "mechanism": [ - "adversarial-verification", - "independent-reimplementation", - "symbolic-trig-closed-form", - "certified-interval-inequality" - ], - "one_line": "From-scratch re-verification confirms 003 end to end: glide reduction and mod-360 case tree re-derived by hand (1.9M-point adversarial sign search, zero dropped-branch hits), I1-I3 re-proven in an independent ring plus an off-torus grid-interpolation proof for W(2,1), Lemma C re-certified with an independent interval stack (4 members), all 8 exact alive certificates match digit-for-digit, four out-of-sample deaths re-measured to about 1e-12, mirror-half windows confirmed; two wording-level corrections, nothing load-bearing refuted.", - "leak_terms": [ - "dlsk_", - "off-torus grid", - "glide action p -> 2m - p", - "1938477" - ], - "range": "identities and glide facts re-proven for all 15 members (own ring; proof-grade off-torus grid for W(2,1); 20 exact spot points each); Lemma C re-certified for (2,1),(5,3),(10,9),(12,12); case tree searched at 1.94M adversarial points across all 15 members; all 8 exact alive certificates re-verified; deaths re-measured for (10,9),(12,12),(5,3),(4,2) plus calibration (2,1),(3,3); dead hunts at gamma_d + {0..0.5} deg for 4 members", - "verifies": "003", - "corrections": [ - "the W(5,3) exact alive certificate is at gamma_d - 1e-6, not the gamma_d - 1e-8 stated in 003 section on the general law (the Outcome/range/Death-brackets statements are correct; only that one sentence is wrong)", - "003 Outcome bullet corridor is empty at every gamma >= gamma_d overstates the theorem: it proves no positive width; a zero-width touching corridor at gamma_d is not excluded (alive is defined as positive width throughout, so nothing downstream is damaged)" - ], - "refuted_by": null, - "gaps": [] - }, - { - "id": "005", - "file": "attempts/005-complete-death-law-theorem.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED", - "mechanism": [ - "formal-parameter-specialization", - "symbolic-trig-closed-form", - "certified-interval-inequality", - "unfolding-word-census", - "out-of-sample-prediction" - ], - "one_line": "Death law completed on both sides: the half word composes to R_0 Rot_A(2a.alpha) Rot_B(-2b.beta), so I1-I4 and the glide facts become one exact 4-variable Laurent check valid for ALL (a,b); Lemmas C and D fall to an elementary c.cot(cs) monotonicity; the case tree is closed for all a >= b >= 1, a >= 2 (the a <= 2b+3 restriction removed); and certified alive segments into the death corner give death(W(a,b)) = gamma_d(a,b) EXACTLY (sup not attained) for 20 members including the newly measured W(6,3), W(6,1), W(8,2), W(8,1), W(7,1).", - "leak_terms": [ - "plaw_", - "R_0 Rot_A(2a alpha) Rot_B(-2b beta)", - "identity I4", - "margin N4", - "c cot(cs) monotone", - "Lemma L1", - "aw - bv = a(b+1)(theta - theta_d) - 360kb", - "rho = 2, t0 = 1/4", - "death corner (90/a, 90(a-1)/(a(b+1)))", - "585/4", - "255/2", - "285/2", - "1035/8", - "900/7", - "three-fold degenerate corner" - ], - "range": "I1-I4, D1, D2 and glide facts proven for ALL integers a, b >= 1 (formal ring; specialization cross-checked exactly on 23 members, floats to a = 40); Lemma C proven for all a >= 2, 1 <= b <= a and Lemma D for all b >= 1; necessity theorem for all integers a >= b >= 1, a >= 2 (no upper restriction on a); exact death = gamma_d certified for 20 members via alive segments t in (0, 1/4], gamma sweeping [gamma_d - 1/4, gamma_d); new float deaths for (6,3),(6,1),(8,2),(8,1),(7,1) at gamma_d to ~5e-12; adversarial case-tree scans ~1.56M points over 13 members", - "gaps": [ - "parametric-sufficiency-open", - "a-equals-1-column-unmeasured", - "birth-side-unproven" - ] - }, - { - "id": "006", - "file": "attempts/006-design-family-past-135.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED", - "mechanism": [ - "symbolic-trig-closed-form", - "unfolding-word-census", - "exact-rational-arithmetic", - "out-of-sample-prediction", - "structure-class-design-search" - ], - "one_line": "The 135 deg stall dissolves: 001 pinch gap was a sampler artifact (W(4,3) certified alive at gamma in (135.0000001, 135.048)), births obey a closed form gamma_birth = 180 - 90(a+b+1)/(a(b+1)) making staircase windows touch, the never-measured W(5,2) and a genuinely new four-block word are certified alive at gamma EXACTLY 135 (rational apexes on the arc, exact orbits), and the W family covers all 157 sampled obtuse arcs - while the design screens (3000+ canonical doubled words to length 50) find nothing of length <= 26 alive past 135 and show clean angle laws are confined to W's two-fan factorization.", - "leak_terms": [ - "design_", - "gamma_birth = 180 - 90(a+b+1)/(a(b+1))", - "W(5,2)", - "0(12)^3(02)^3(12)^2(02)^3", - "exactly 135", - "windows touch", - "(2x-1)^2 + (2y+1)^2 = 2", - "edge-accumulating sampler", - "136.8265" - ], - "range": "five exact certificates (rational apexes, Fraction corridors, independently simulated orbits): W(4,3) at certified gamma in (135.001,135.048) and (135.0000001,135.001); W(5,4) at (144.0000001,144.001); W(5,2) len 30 and non-W four-block N1 len 46 at gamma exactly 135 (exact on-arc identity); 003 obligations machine-certified for two new members (5,2),(6,2). Float screens sample-bounded: 245 canonical doubled words len <= 26 dead at 13 arcs >= 135; 811 len-30 words (only W(4,3),W(5,2) alive on 135-144); 2008 two-alternating words len 34-50 (only W members + two four-block words); three-block family a,b,c <= 3 (25/27 dead everywhere obtuse); birth law float-checked at 11 members; coverage checked at 157 arcs 90.5-165", - "gaps": [ - "birth-law-unproven (gamma_birth closed form is float-fitted + factor-structure motivated; no necessity/sufficiency theorem)", - "coverage-conjecture-open (union of windows covering (90,180) reduces to an elementary Diophantine statement, unproven)", - "length-26-stall-sample-bounded (doubled words only, 13 arcs; non-doubled even words 24-26 not rescreened)", - "four-block-death-angles-uncharacterized (binding differences provably non-elementary here only by failed factor search)" - ] - }, - { - "id": "007", - "file": "attempts/007-skeptic-review-of-006.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED_WITH_CORRECTIONS", - "mechanism": [ - "adversarial-verification", - "independent-reimplementation", - "exact-rational-arithmetic", - "out-of-sample-prediction" - ], - "one_line": "From-scratch re-verification confirms 006 everywhere it is exact: all five certificates re-established by a third corridor implementation (digit-for-digit widths) plus an own exact simulator, the exactly-135 on-arc identity re-derived by hand, 001's pinch-gap birth value regenerated from a replica of its own sampler (mechanism: the alive window is a corner sliver inside the grid's terminal 6e-4 clearance), the birth law survives three genuinely out-of-sample members to ~3e-11 (= sampler floor, including a > 2b+3), and the (5,2)/(6,2) theorem extension re-proves in an independent ring; four corrections, none load-bearing.", - "leak_terms": [ - "dsk_", - "566 not 811", - "canonical pairing (a,b) ~ (b+1,a-1)", - "corner sliver 6e-4", - "floor-tracking", - "W(8,2) birth 138.75" - ], - "range": "all 5 exact certificates re-verified (own matrix-form exact corridor, widths match digit-for-digit; own exact orbit simulation; own certified gamma brackets via 004's interval stack; on-arc gamma=135 identity hand-derived and Fraction-checked; Niven applicability checked); W(4,3) birth floor-tracked at 4 sampler depths; 001's census sampler replicated (reproduces 135.0486) and window x-extent measured at 7 gammas; birth law tested out-of-sample on W(7,3), W(6,3), W(8,2) (all +3.0e-11 = floor, dead-sampled below); coverage re-run at 25 own arcs (25/25 alive); universes recounted (245 and 2008 confirmed, 566-vs-811 correction); s1 hits reclassified (14 W + N1/N2); N1/N2 non-W checked incl. 0-1 swap; 60-word random dead re-test + 19 near-miss attacks (zero kills); identities + case tree (550k points) + Lemma C re-certified for (5,2),(6,2). NOT re-swept: full length-30 and s1 negative universes, three-block table, design_factor.py factorizations", - "verifies": "006", - "corrections": [ - "universe counts: the length-30 (|u|=15) universe is 566 canonical words, not 811 (811 is |u| <= 15 cumulative); 'all doubled odd words to length 30: 1056' double-counts (true count 811); the index one_line's '3000+' is 2846 counted correctly - screen coverage itself is unaffected", - "'length 30 is alive throughout the measured gap' overstates: certificates prove alive points AT gamma = 135 exactly and at one gamma in each of (135.0000001, 135.001) and (135.001, 135.048); the continuum statement is float evidence plus the SPECULATION window law", - "refutation semantics vs 001: 001's gap claim was knowledge-scoped ('nothing <= 30 KNOWN alive') and stays true as written; what 006 overturns is 001's birth measurements (error 0.0486 deg, 10x the stated ~0.005 sampling noise, whose 'deaths only under-estimated' hedge does not transfer to births) - correction/closure, not refutation", - "the window pairing (a,b) <-> (b+1, a-1) is a canonical-word identity (0<->1 swap + rotation/reversal; verified exactly on six pairs), so it is vacuous as evidence for the birth law, 006's 11-member fit spans only 9 canonical words, and 003's 'two distinct words dying on the same arc' (W(3,3)/W(4,2)) is a relabeling of one canonical word" - ], - "refuted_by": null, - "gaps": [] - }, - { - "id": "008", - "file": "attempts/008-skeptic-review-of-005.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED_WITH_CORRECTIONS", - "mechanism": [ - "adversarial-verification", - "independent-reimplementation", - "formal-parameter-specialization", - "symbolic-trig-closed-form", - "certified-interval-inequality" - ], - "one_line": "Adversarial re-verification confirms 005 end to end: the closed-form bridge u = R_0 Rot_A(2a.alpha) Rot_B(-2b.beta) recomposed exactly at off-torus rational points (40 members to a = 40), I1-I4 and glide facts re-proven in an independent 4-variable ring with u re-derived by symbolic map composition, Lemmas L1/C/D re-derived by hand and stressed to a = 10^6, the extended case tree re-derived branch by branch (threshold identity exact; 447k-point scan on 10 a > 2b+3 members, zero hits), the sufficiency segments independently certified alive by full-corridor interval unfolding down to gamma_d - 9.3e-10, and all five new deaths re-measured on the mirror half; one numeric correction, nothing load-bearing refuted.", - "leak_terms": [ - "psk_", - "off-torus pair algebra", - "gamma_d - 9.3e-10", - "447k", - "1e-4 not 1e-6" - ], - "range": "bridge recomposed exactly for 40 members incl. (40,40) at 2 rational points each; identities re-proven formally (my ring, 14/14); Lemma numerics 14,183 checks to a = 10^6; case tree scanned at ~447k adversarial points over 10 members with a > 2b+3 (k up to 2 reachable); segments certified alive at 42 points (7 members incl. the 5 new, t down to 2^-30); plaw_suffice trig layer audited (596 enclosure checks); 5 deaths re-measured with 004's sampler; plaw_suffice all NOT re-run end to end (selftest + one member reproduced, code audited)", - "verifies": "005", - "corrections": [ - "005 states the five new exact-alive certificates certify gamma within 1e-6 of gamma_d; the data files show gamma_d - 1e-4 for W(6,3), W(8,2), W(8,1), W(7,1) (only W(6,1) reaches 1e-6) - the --dg 1e-6 request is a float target, the certified bound is the largest 10^-k that certifies after apex snapping", - "caveat, not an error: the five new members' necessity rests on the parametric theorem alone (no per-member deathlaw_prove obligations were run for them, unlike 003's fifteen); sound by design since the identities are now proven for all (a,b)" - ], - "refuted_by": null, - "gaps": [] - }, - { - "id": "009", - "file": "attempts/009-lean-pilot-cot-lemma.md", - "date": "2026-08-04", - "mode": "informed", - "status": "FORMALIZED", - "mechanism": [ - "machine-checked-formalization", - "symbolic-trig-closed-form" - ], - "one_line": "First Lean 4 certificate in the lab: Lemma L1 of 005 (c.cot(cs) strictly decreasing on (0,1] for fixed s in (0, pi/2]) proved in mathlib with zero sorries and only the three standard axioms, statement matching 005 hypothesis-by-hypothesis with no narrowing; independent statement review and independent rebuild still pending.", - "leak_terms": [ - "cot_lemma_L1", - "strictAntiOn_id_mul_cot", - "BilliardsFormal", - "no HasDerivAt lemma for Real.cot", - "first Lean certificate" - ], - "verifies": "005", - "range": "exactly the formal statement StrictAntiOn (fun c => c * Real.cot (c * s)) (Ioc 0 1) for all real s with 0 < s <= pi/2 (Lean 4.32.2, mathlib v4.32.2); Lemmas C and D themselves, and the I1-I4 Laurent block, are NOT formalized", - "gaps": [ - "laurent-block-I1-I4-not-attempted (queue item 16 second half)" - ] - }, - { - "id": "010", - "file": "attempts/010-statement-review-of-009.md", - "date": "2026-08-04", - "mode": "informed", - "status": "VERIFIED", - "mechanism": [ - "adversarial-verification", - "machine-checked-formalization" - ], - "one_line": "Independent completion of the FORMALIZE.md pass on 009: non-formalizer statement review finds cot_lemma_L1 matches 005's Lemma L1 hypothesis-for-hypothesis with no narrowing (StrictAntiOn/Ioc/Real.cot read from vendored mathlib source, no junk-value reliance, cheat scan clean), own fresh lake build after deleting the project build output is green with axioms exactly [propext, Classical.choice, Quot.sound], and all three of 005's L1 use sites instantiate inside the formalized hypotheses; zero corrections, two cosmetic nits.", - "leak_terms": [ - "SkepticAxiomCheck", - "statement review signed off", - "AxiomCheck.lean not committed" - ], - "verifies": "009", - "range": "statement fidelity vs 005 section (ii)-B checked hypothesis by hypothesis; definitions Real.cot, StrictAntiOn, Set.Ioc read from mathlib rev 905b958 source; cheat scan over both project .lean files; rebuild = rm -rf .lake/build + lake build (8657 jobs, green) in a fresh container session with the pinned toolchain, mathlib oleans from the existing pinned cache (not re-downloaded); own axiom audit of both theorems; usage check of all three L1 call sites in 005. NOT audited: mathlib internals beyond the definitions used, Lean kernel, Lemmas C/D themselves", - "gaps": [] - } - ] -} + "problem": "billiards-triangles", + "route_status": "EVIDENCE", + "route_summary": "Word-census route mapped blind to length 26: 376 exact certificates (family orbits to length 66, gamma <= 160 deg), coverage staircase stalling at 135 deg, and a measured hyperbolic (not super-exponential) growth of needed word length vs angle - but certified box width collapses exponentially in length, so length buys angle reach, not area. Negatives are sample-bounded; skeptic-confirmed with one numeric correction.", + "attempts": [ + { + "id": "001", + "file": "attempts/001-word-census-coverage-map.md", + "date": "2026-07-30", + "mode": "blind", + "status": "EVIDENCE", + "mechanism": [ + "unfolding-word-census", + "certified-box-coverage", + "exhaustive-enumeration" + ], + "one_line": "Canonical word census to length 26 with 376 exact certificates: coverage staircase stalls at 135 deg, blind rediscovery of the ~112.5 deg frontier, and a hyperbolic certificate-length growth law along the W(a,b) family (checked to length 66).", + "leak_terms": [ + "112.4989", + "W(a,b)", + "coverage staircase", + "length buys angle reach", + "17,527" + ], + "range": "canonical translation words of even length <= 26 (17,527 words); 1/64-cell midpoint screen of the obtuse half-disk; exact dyadic-box anchors; family rows to W(8,8), gamma <= 160 deg", + "gaps": [ + "sample-bounded-negatives", + "death-angle-laws-unproven" + ] + }, + { + "id": "002", + "file": "attempts/002-skeptic-review-of-001.md", + "date": "2026-07-30", + "mode": "blind", + "status": "VERIFIED_WITH_CORRECTIONS", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "unfolding-word-census" + ], + "one_line": "From-scratch re-implementation of every layer confirms 001 (36/36 sampled certificates at 180 exact apexes, 17,527-word universe reproduced word-for-word, death law survives two out-of-sample prediction rows); certified-area figure corrected to ~4.3e-6.", + "leak_terms": [ + "147.5974", + "157.4976", + "4.3e-6" + ], + "verifies": "001", + "corrections": [ + "certified-area total over all 376 anchors is ~4.3e-6, not ~2.2e-6 (which is the map-anchor subtotal)", + "the 64.3% uncovered figure is of the resolved interior, not the whole obtuse region", + "anchor simulation re-derivation is at the anchor centre; the full 9-sample box attack ran on every 5th/7th record", + "kill-condition verdict is measured only to gamma ~ 160 deg; beyond that the hyperbolic form is the SPECULATION law's extrapolation" + ] + }, + { + "id": "003", + "file": "attempts/003-death-angle-laws.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED", + "mechanism": [ + "symbolic-trig-closed-form", + "certified-interval-inequality", + "out-of-sample-prediction", + "unfolding-word-census" + ], + "one_line": "The W(a,b) death-angle laws of 001 survive both out-of-sample tests (W(10,9) -> 162.90, W(12,12) -> 2160/13, to ~1e-12), and are then derived: exact division-free unfolding gives closed-form binding functions, a general law gamma_d = 180 - 90(a+b)/(a(b+1)) (confirmed out-of-sample on W(5,3) -> 144 and W(4,2) -> 135), and a machine-certified necessity theorem: the corridor is empty at every gamma >= gamma_d, for 15 verified members.", + "leak_terms": [ + "gamma_d = 180 - 90(a+b)/(a(b+1))", + "90(a+b)/(a(b+1))", + "glide reflection reduction", + "W(5,3)", + "144.0000000000", + "deathlaw_", + "Lemma C", + "division-free unfolding" + ], + "range": "necessity theorem machine-certified for the 15 members (2,1),(2,2),(3,2),(3,3),(4,2),(4,3),(4,4),(5,3),(5,4),(5,5),(6,6),(7,7),(8,8),(10,9),(12,12); exact alive certificates at gamma_d - 1e-6 (1e-8 for (3,3),(4,2)) for 8 members; float deaths measured to ~1e-11 for 11 members; parametric statement for all a >= b >= 1, a >= 2, a <= 2b+3 rests on two per-member-verified inputs (identities I1-I3 and Lemma C) labelled SPECULATION in general", + "gaps": [ + "sufficiency-limit-unclosed (death = gamma_d exactly needs alive points approaching gamma_d; certified alive stops at gamma_d - 1e-6)", + "parametric-identities-unproven-in-general (I1-I3 and Lemma C verified per member, 15 members; general (a,b) statement labelled SPECULATION)", + "a-greater-than-2b+3-uncovered (branch analysis breaks; W(6,3)-type members unmeasured)" + ], + "refuted_by": null + }, + { + "id": "004", + "file": "attempts/004-skeptic-review-of-003.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED_WITH_CORRECTIONS", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "symbolic-trig-closed-form", + "certified-interval-inequality" + ], + "one_line": "From-scratch re-verification confirms 003 end to end: glide reduction and mod-360 case tree re-derived by hand (1.9M-point adversarial sign search, zero dropped-branch hits), I1-I3 re-proven in an independent ring plus an off-torus grid-interpolation proof for W(2,1), Lemma C re-certified with an independent interval stack (4 members), all 8 exact alive certificates match digit-for-digit, four out-of-sample deaths re-measured to about 1e-12, mirror-half windows confirmed; two wording-level corrections, nothing load-bearing refuted.", + "leak_terms": [ + "dlsk_", + "off-torus grid", + "glide action p -> 2m - p", + "1938477" + ], + "range": "identities and glide facts re-proven for all 15 members (own ring; proof-grade off-torus grid for W(2,1); 20 exact spot points each); Lemma C re-certified for (2,1),(5,3),(10,9),(12,12); case tree searched at 1.94M adversarial points across all 15 members; all 8 exact alive certificates re-verified; deaths re-measured for (10,9),(12,12),(5,3),(4,2) plus calibration (2,1),(3,3); dead hunts at gamma_d + {0..0.5} deg for 4 members", + "verifies": "003", + "corrections": [ + "the W(5,3) exact alive certificate is at gamma_d - 1e-6, not the gamma_d - 1e-8 stated in 003 section on the general law (the Outcome/range/Death-brackets statements are correct; only that one sentence is wrong)", + "003 Outcome bullet corridor is empty at every gamma >= gamma_d overstates the theorem: it proves no positive width; a zero-width touching corridor at gamma_d is not excluded (alive is defined as positive width throughout, so nothing downstream is damaged)" + ], + "refuted_by": null, + "gaps": [] + }, + { + "id": "005", + "file": "attempts/005-complete-death-law-theorem.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED", + "mechanism": [ + "formal-parameter-specialization", + "symbolic-trig-closed-form", + "certified-interval-inequality", + "unfolding-word-census", + "out-of-sample-prediction" + ], + "one_line": "Death law completed on both sides: the half word composes to R_0 Rot_A(2a.alpha) Rot_B(-2b.beta), so I1-I4 and the glide facts become one exact 4-variable Laurent check valid for ALL (a,b); Lemmas C and D fall to an elementary c.cot(cs) monotonicity; the case tree is closed for all a >= b >= 1, a >= 2 (the a <= 2b+3 restriction removed); and certified alive segments into the death corner give death(W(a,b)) = gamma_d(a,b) EXACTLY (sup not attained) for 20 members including the newly measured W(6,3), W(6,1), W(8,2), W(8,1), W(7,1).", + "leak_terms": [ + "plaw_", + "R_0 Rot_A(2a alpha) Rot_B(-2b beta)", + "identity I4", + "margin N4", + "c cot(cs) monotone", + "Lemma L1", + "aw - bv = a(b+1)(theta - theta_d) - 360kb", + "rho = 2, t0 = 1/4", + "death corner (90/a, 90(a-1)/(a(b+1)))", + "585/4", + "255/2", + "285/2", + "1035/8", + "900/7", + "three-fold degenerate corner" + ], + "range": "I1-I4, D1, D2 and glide facts proven for ALL integers a, b >= 1 (formal ring; specialization cross-checked exactly on 23 members, floats to a = 40); Lemma C proven for all a >= 2, 1 <= b <= a and Lemma D for all b >= 1; necessity theorem for all integers a >= b >= 1, a >= 2 (no upper restriction on a); exact death = gamma_d certified for 20 members via alive segments t in (0, 1/4], gamma sweeping [gamma_d - 1/4, gamma_d); new float deaths for (6,3),(6,1),(8,2),(8,1),(7,1) at gamma_d to ~5e-12; adversarial case-tree scans ~1.56M points over 13 members", + "gaps": [ + "parametric-sufficiency-open", + "a-equals-1-column-unmeasured", + "birth-side-unproven" + ] + }, + { + "id": "006", + "file": "attempts/006-design-family-past-135.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED", + "mechanism": [ + "symbolic-trig-closed-form", + "unfolding-word-census", + "exact-rational-arithmetic", + "out-of-sample-prediction", + "structure-class-design-search" + ], + "one_line": "The 135 deg stall dissolves: 001 pinch gap was a sampler artifact (W(4,3) certified alive at gamma in (135.0000001, 135.048)), births obey a closed form gamma_birth = 180 - 90(a+b+1)/(a(b+1)) making staircase windows touch, the never-measured W(5,2) and a genuinely new four-block word are certified alive at gamma EXACTLY 135 (rational apexes on the arc, exact orbits), and the W family covers all 157 sampled obtuse arcs - while the design screens (3000+ canonical doubled words to length 50) find nothing of length <= 26 alive past 135 and show clean angle laws are confined to W's two-fan factorization.", + "leak_terms": [ + "design_", + "gamma_birth = 180 - 90(a+b+1)/(a(b+1))", + "W(5,2)", + "0(12)^3(02)^3(12)^2(02)^3", + "exactly 135", + "windows touch", + "(2x-1)^2 + (2y+1)^2 = 2", + "edge-accumulating sampler", + "136.8265" + ], + "range": "five exact certificates (rational apexes, Fraction corridors, independently simulated orbits): W(4,3) at certified gamma in (135.001,135.048) and (135.0000001,135.001); W(5,4) at (144.0000001,144.001); W(5,2) len 30 and non-W four-block N1 len 46 at gamma exactly 135 (exact on-arc identity); 003 obligations machine-certified for two new members (5,2),(6,2). Float screens sample-bounded: 245 canonical doubled words len <= 26 dead at 13 arcs >= 135; 811 len-30 words (only W(4,3),W(5,2) alive on 135-144); 2008 two-alternating words len 34-50 (only W members + two four-block words); three-block family a,b,c <= 3 (25/27 dead everywhere obtuse); birth law float-checked at 11 members; coverage checked at 157 arcs 90.5-165", + "gaps": [ + "birth-law-unproven (gamma_birth closed form is float-fitted + factor-structure motivated; no necessity/sufficiency theorem)", + "coverage-conjecture-open (union of windows covering (90,180) reduces to an elementary Diophantine statement, unproven)", + "length-26-stall-sample-bounded (doubled words only, 13 arcs; non-doubled even words 24-26 not rescreened)", + "four-block-death-angles-uncharacterized (binding differences provably non-elementary here only by failed factor search)" + ] + }, + { + "id": "007", + "file": "attempts/007-skeptic-review-of-006.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED_WITH_CORRECTIONS", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "exact-rational-arithmetic", + "out-of-sample-prediction" + ], + "one_line": "From-scratch re-verification confirms 006 everywhere it is exact: all five certificates re-established by a third corridor implementation (digit-for-digit widths) plus an own exact simulator, the exactly-135 on-arc identity re-derived by hand, 001's pinch-gap birth value regenerated from a replica of its own sampler (mechanism: the alive window is a corner sliver inside the grid's terminal 6e-4 clearance), the birth law survives three genuinely out-of-sample members to ~3e-11 (= sampler floor, including a > 2b+3), and the (5,2)/(6,2) theorem extension re-proves in an independent ring; four corrections, none load-bearing.", + "leak_terms": [ + "dsk_", + "566 not 811", + "canonical pairing (a,b) ~ (b+1,a-1)", + "corner sliver 6e-4", + "floor-tracking", + "W(8,2) birth 138.75" + ], + "range": "all 5 exact certificates re-verified (own matrix-form exact corridor, widths match digit-for-digit; own exact orbit simulation; own certified gamma brackets via 004's interval stack; on-arc gamma=135 identity hand-derived and Fraction-checked; Niven applicability checked); W(4,3) birth floor-tracked at 4 sampler depths; 001's census sampler replicated (reproduces 135.0486) and window x-extent measured at 7 gammas; birth law tested out-of-sample on W(7,3), W(6,3), W(8,2) (all +3.0e-11 = floor, dead-sampled below); coverage re-run at 25 own arcs (25/25 alive); universes recounted (245 and 2008 confirmed, 566-vs-811 correction); s1 hits reclassified (14 W + N1/N2); N1/N2 non-W checked incl. 0-1 swap; 60-word random dead re-test + 19 near-miss attacks (zero kills); identities + case tree (550k points) + Lemma C re-certified for (5,2),(6,2). NOT re-swept: full length-30 and s1 negative universes, three-block table, design_factor.py factorizations", + "verifies": "006", + "corrections": [ + "universe counts: the length-30 (|u|=15) universe is 566 canonical words, not 811 (811 is |u| <= 15 cumulative); 'all doubled odd words to length 30: 1056' double-counts (true count 811); the index one_line's '3000+' is 2846 counted correctly - screen coverage itself is unaffected", + "'length 30 is alive throughout the measured gap' overstates: certificates prove alive points AT gamma = 135 exactly and at one gamma in each of (135.0000001, 135.001) and (135.001, 135.048); the continuum statement is float evidence plus the SPECULATION window law", + "refutation semantics vs 001: 001's gap claim was knowledge-scoped ('nothing <= 30 KNOWN alive') and stays true as written; what 006 overturns is 001's birth measurements (error 0.0486 deg, 10x the stated ~0.005 sampling noise, whose 'deaths only under-estimated' hedge does not transfer to births) - correction/closure, not refutation", + "the window pairing (a,b) <-> (b+1, a-1) is a canonical-word identity (0<->1 swap + rotation/reversal; verified exactly on six pairs), so it is vacuous as evidence for the birth law, 006's 11-member fit spans only 9 canonical words, and 003's 'two distinct words dying on the same arc' (W(3,3)/W(4,2)) is a relabeling of one canonical word" + ], + "refuted_by": null, + "gaps": [] + }, + { + "id": "008", + "file": "attempts/008-skeptic-review-of-005.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED_WITH_CORRECTIONS", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "formal-parameter-specialization", + "symbolic-trig-closed-form", + "certified-interval-inequality" + ], + "one_line": "Adversarial re-verification confirms 005 end to end: the closed-form bridge u = R_0 Rot_A(2a.alpha) Rot_B(-2b.beta) recomposed exactly at off-torus rational points (40 members to a = 40), I1-I4 and glide facts re-proven in an independent 4-variable ring with u re-derived by symbolic map composition, Lemmas L1/C/D re-derived by hand and stressed to a = 10^6, the extended case tree re-derived branch by branch (threshold identity exact; 447k-point scan on 10 a > 2b+3 members, zero hits), the sufficiency segments independently certified alive by full-corridor interval unfolding down to gamma_d - 9.3e-10, and all five new deaths re-measured on the mirror half; one numeric correction, nothing load-bearing refuted.", + "leak_terms": [ + "psk_", + "off-torus pair algebra", + "gamma_d - 9.3e-10", + "447k", + "1e-4 not 1e-6" + ], + "range": "bridge recomposed exactly for 40 members incl. (40,40) at 2 rational points each; identities re-proven formally (my ring, 14/14); Lemma numerics 14,183 checks to a = 10^6; case tree scanned at ~447k adversarial points over 10 members with a > 2b+3 (k up to 2 reachable); segments certified alive at 42 points (7 members incl. the 5 new, t down to 2^-30); plaw_suffice trig layer audited (596 enclosure checks); 5 deaths re-measured with 004's sampler; plaw_suffice all NOT re-run end to end (selftest + one member reproduced, code audited)", + "verifies": "005", + "corrections": [ + "005 states the five new exact-alive certificates certify gamma within 1e-6 of gamma_d; the data files show gamma_d - 1e-4 for W(6,3), W(8,2), W(8,1), W(7,1) (only W(6,1) reaches 1e-6) - the --dg 1e-6 request is a float target, the certified bound is the largest 10^-k that certifies after apex snapping", + "caveat, not an error: the five new members' necessity rests on the parametric theorem alone (no per-member deathlaw_prove obligations were run for them, unlike 003's fifteen); sound by design since the identities are now proven for all (a,b)" + ], + "refuted_by": null, + "gaps": [] + }, + { + "id": "009", + "file": "attempts/009-lean-pilot-cot-lemma.md", + "date": "2026-08-04", + "mode": "informed", + "status": "FORMALIZED", + "mechanism": [ + "machine-checked-formalization", + "symbolic-trig-closed-form" + ], + "one_line": "First Lean 4 certificate in the lab: Lemma L1 of 005 (c.cot(cs) strictly decreasing on (0,1] for fixed s in (0, pi/2]) proved in mathlib with zero sorries and only the three standard axioms, statement matching 005 hypothesis-by-hypothesis with no narrowing; independent statement review and independent rebuild still pending.", + "leak_terms": [ + "cot_lemma_L1", + "strictAntiOn_id_mul_cot", + "BilliardsFormal", + "no HasDerivAt lemma for Real.cot", + "first Lean certificate" + ], + "verifies": "005", + "range": "exactly the formal statement StrictAntiOn (fun c => c * Real.cot (c * s)) (Ioc 0 1) for all real s with 0 < s <= pi/2 (Lean 4.32.2, mathlib v4.32.2); Lemmas C and D themselves, and the I1-I4 Laurent block, are NOT formalized", + "gaps": [ + "laurent-block-I1-I4-not-attempted (queue item 16 second half)" + ] + }, + { + "id": "010", + "file": "attempts/010-statement-review-of-009.md", + "date": "2026-08-04", + "mode": "informed", + "status": "VERIFIED", + "mechanism": [ + "adversarial-verification", + "machine-checked-formalization" + ], + "one_line": "Independent completion of the FORMALIZE.md pass on 009: non-formalizer statement review finds cot_lemma_L1 matches 005's Lemma L1 hypothesis-for-hypothesis with no narrowing (StrictAntiOn/Ioc/Real.cot read from vendored mathlib source, no junk-value reliance, cheat scan clean), own fresh lake build after deleting the project build output is green with axioms exactly [propext, Classical.choice, Quot.sound], and all three of 005's L1 use sites instantiate inside the formalized hypotheses; zero corrections, two cosmetic nits.", + "leak_terms": [ + "SkepticAxiomCheck", + "statement review signed off", + "AxiomCheck.lean not committed" + ], + "verifies": "009", + "range": "statement fidelity vs 005 section (ii)-B checked hypothesis by hypothesis; definitions Real.cot, StrictAntiOn, Set.Ioc read from mathlib rev 905b958 source; cheat scan over both project .lean files; rebuild = rm -rf .lake/build + lake build (8657 jobs, green) in a fresh container session with the pinned toolchain, mathlib oleans from the existing pinned cache (not re-downloaded); own axiom audit of both theorems; usage check of all three L1 call sites in 005. NOT audited: mathlib internals beyond the definitions used, Lean kernel, Lemmas C/D themselves", + "gaps": [] + }, + { + "id": "011", + "file": "attempts/011-lean-laurent-block.md", + "date": "2026-08-04", + "mode": "informed", + "status": "FORMALIZED", + "mechanism": [ + "machine-checked-formalization", + "formal-parameter-specialization" + ], + "one_line": "Second Lean 4 certificate: the full Laurent-identity block of 005 - closed-form half-word composition R0.RotA(2a alpha).RotB(-2b beta), I1-I4, D1, D2 and the glide facts - kernel-checked with (a,b) as formal exponent variables (23 theorems, zero sorries, standard three axioms only), tiers A+B+C green plus the letter-word iterate version for all natural (a,b); the block turns out homogeneous in i (no i^2 = -1 needed); geometry bridge out of scope per the lane, statement review pending.", + "leak_terms": [ + "LaurentBlock", + "half_word_letters", + "half_word_closed_form", + "homogeneous in i", + "pair encoding conj substitution", + "no AddMonoidAlgebra needed" + ], + "verifies": "005", + "range": "exactly the formal statements in formal/BilliardsFormal/LaurentBlock.lean: the identity block over an arbitrary field K with nonzero generators x, y, P, Q, i and 2 != 0 (P, Q carrying the (a,b) exponents formally; Tier C specializes P = x^a, Q = y^b for all integers a, b; half_word_letters covers the genuine letter word for all natural a, b); Lean 4.32.2, mathlib v4.32.2, axioms exactly [propext, Classical.choice, Quot.sound]", + "gaps": [ + "pending-independent-statement-review (encoding table vs 005 (ii)-A; FORMALIZE.md part 1)", + "pending-independent-rebuild (FORMALIZE.md part 3)", + "geometry-bridge-not-formalized (ring model = unfolding geometry carried by 005/008, out of scope by design)" + ] + } + ] +} \ No newline at end of file From e852ebbbdf523e7c1d3e0ac5b5361ed56db79679 Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 10:34:15 +0000 Subject: [PATCH 08/14] billiards-triangles 012: statement review of 011 - Laurent-block FORMALIZED pass complete Independent skeptic pass, zero corrections: all 23 theorems mapped to 005 statement-by-statement (angle monomials, factor-of-2 conventions, signs and orientations all verbatim; permutation/sign-swap hunts came up empty), conjugation-as-substitution confirmed as 008's star involution with nothing smuggled (key theorems quantify over arbitrary pairs), dropped i^2=-1 confirmed safe-direction and independently re-proved with i as a formal variable, N/Z parameter coverage confirmed non-narrowing, cheat scan clean, independent rebuild green with own axiom audit. All 31 identities re-derived exactly in a from-scratch stdlib Laurent-ring engine (lbsk_review.py). Geometry bridge remains permanent out-of-scope carried by 005/008. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- .../attempts/012-statement-review-of-011.md | 402 +++++++++++ .../billiards-triangles/data/lbsk_review.json | 72 ++ .../explore/lbsk_review.py | 387 ++++++++++ problems/billiards-triangles/prior-art.json | 660 +++++++++--------- 4 files changed, 1202 insertions(+), 319 deletions(-) create mode 100644 problems/billiards-triangles/attempts/012-statement-review-of-011.md create mode 100644 problems/billiards-triangles/data/lbsk_review.json create mode 100644 problems/billiards-triangles/explore/lbsk_review.py diff --git a/problems/billiards-triangles/attempts/012-statement-review-of-011.md b/problems/billiards-triangles/attempts/012-statement-review-of-011.md new file mode 100644 index 0000000..0731335 --- /dev/null +++ b/problems/billiards-triangles/attempts/012-statement-review-of-011.md @@ -0,0 +1,402 @@ +# 012 — Statement review + independent rebuild of 011 (Lean Laurent block) + +- **Problem:** billiards-triangles, `problems/billiards-triangles/PROBLEM.md` +- **Date:** 2026-08-04 +- **Mode:** informed (read `prior-art.json`, 011 in full, 005 in full — + the source of truth for the statements — 008 §§1–2 (the second witness: + star involution, off-torus pair algebra, v₂'s geometric identification), + `docs/FORMALIZE.md`, 010 for the pass shape, + `explore/plaw_general.py` line by line (`closed_forms`, `cmd_formal` — + the proof object 011 transcribed), and every line of the five files + under `formal/` outside `.lake/`) +- **Type:** skeptic review of `011-lean-laurent-block.md` (default stance: + REFUTE). Independent completion of the three-part `FORMALIZED` pass of + `docs/FORMALIZE.md` for 005's Laurent-identity block: (1) statement + review by a non-formalizer — the main event here, since the encoding is + nontrivial and a wrong mapping is invisible to the kernel — (2) clean + build, (3) reproduction via my own `lake build`. I wrote none of the + Lean code under review. +- **Outcome in one line:** 011 survives on every attacked front. The 23 + Lean statements are 005's statements under the encoding table, checked + object by object and sign by sign; every theorem re-proven from scratch + in my own 7-variable symbolic ring (i formal — independently confirming + the i-homogeneity claim); the statements at the torus instance match + 005's prose quantities to ~4e-15 at 200 random (a, b, α, β); my own + fresh `lake build` is green and my own axiom audit of all 23 theorems + shows exactly `[propext, Classical.choice, Quot.sound]`. Zero + corrections, three cosmetic nits. The FORMALIZED pass on the Laurent + block is now **complete** (bridge permanently out of scope, as designed). +- **Tools:** new `explore/lbsk_review.py` (stdlib-only, deterministic, + seed 20260804, ~5 s; output `data/lbsk_review.json`): Layer 1 is a + from-scratch 7-variable Laurent-polynomial ring over ℚ — variables + x, y, P, Q, **i**, z1, z2, so `i` is a formal unit exactly as in the + Lean statements — re-proving every identity by exact polynomial + subtraction, with the definitions transcribed from `LaurentBlock.lean` + (NOT from `plaw_general.py`, which it never imports); Layer 2 + instantiates the Lean expressions at x = e^{iα}, y = e^{iβ}, + P = e^{iaα}, Q = e^{ibβ}, i = 1j and compares against 005's own prose + formulas. Rebuild: elan / Lean `leanprover/lean4:v4.32.2` via + `$HOME/.elan/bin`, `rm -rf .lake/build && lake build` (packages cache + kept, per the lane), then my own scratch `SkepticAxiomCheck011.lean` + (deleted after the run; contents = `#print axioms` on all 23 theorems, + enumerated by me from the source file, not copied from + `AxiomCheck.lean`). +- **Sources:** repo + the pinned mathlib checkout only; no external + papers. + +No file of 011 (or any prior attempt) was modified. + +Reproduce (repo root): + +``` +python3 problems/billiards-triangles/explore/lbsk_review.py --out problems/billiards-triangles/data/lbsk_review.json +cd problems/billiards-triangles/formal && rm -rf .lake/build && lake build # PATH needs $HOME/.elan/bin +# axiom audit: scratch file importing BilliardsFormal with #print axioms on all 23 Laurent theorems +``` + +## Claims attacked + +The assigned list (a)–(h) plus two of my own (i)–(j): + +1. **(a) Statement mapping, theorem by theorem** — that each of the 23 + Lean statements says what 005 §(ii)-A says: variable correspondence, + the sinF/cosF dictionary at the right angles (including the + factor-of-2 conventions in Rot_A(2aα), Rot_B(−2bβ)), sign + conventions, and that `identity_I1..I4`, `D1`, `D2` are 005's I1–I4, + D1, D2 and not renamed/permuted variants. +2. **(b) The conjugation encoding** — conj as the substitution + (x,y,P,Q,i) ↦ (x⁻¹,y⁻¹,P⁻¹,Q⁻¹,−i), points as pairs with + syntactically-substituted second components: does anything in the + block need conj as a ring involution on arbitrary elements; does the + pair encoding smuggle in "second component = star of the first"; is + this 008's star involution? +3. **(c) The dropped i² = −1** — is the generalization direction safe, + i.e. do the statements at the intended instance i = Complex.I say + 005's statements, with sinF/cosF reducing to actual sin/cos there? +4. **(d) Genericity** — (a,b) enter only as exponents of ∀-quantified + generators; the ℕ-iterate + ℤ-specialization combination covers 005's + integer parameters a ≥ b ≥ 1 with no silent narrowing; v₂'s dropped Q + argument is genuinely absent in 005's v₂ too. +5. **(e) Cheat scan** — sorry/admit/native_decide/axiom/unsafe/partial/ + implemented_by/notation shadowing/vacuous hypotheses over all project + .lean files; that the root module imports `LaurentBlock` so the build + elaborates it. +6. **(f) Independent rebuild + axiom audit** — my own `lake build` after + deleting the project build output; my own `#print axioms` on all 23 + theorems from a scratch file I wrote, not the committed + `AxiomCheck.lean`. +7. **(g) Record audit** — does 011 or its index entry overclaim; is the + bridge genuinely excluded in prose; is the numeric i-check labelled a + sanity check; are range/gaps/one_line honest. +8. **(h) Spot re-derivation** — re-derive identities independently of + both 005's prose and the Lean proofs. Done for ALL of them + (Layer 1's from-scratch exact symbolic computation is exactly the + "Python Fractions over a rational function field at symbolic + exponents" route), plus a full by-hand product-to-sum derivation of + D1 as a second, computer-free witness. +9. **(i, own) Coefficient-rationality of the conj convention** — the + substitution is complex conjugation only if every constant in the + definitions is rational (conjugating c(i) ∈ ℚ(i) is i ↦ −i); a single + non-rational literal would silently break the encoding. +10. **(j, own) Toolchain-pin consistency** — `lean-toolchain`, + `lakefile.toml`, `lake-manifest.json`, and the log agree on + Lean 4.32.2 / mathlib rev `905b958…`. + +## Refutations found + +**None.** No statement delta, no sign error, no narrowing, no cheat, no +axiom beyond the standard three, no overclaim. Three cosmetic nits (none +is a correction; nothing downstream changes): + +1. 011's Tools line says `LaurentBlock.lean` is "~420 lines"; the file + is 423 lines. Rounding, noted only for completeness. +2. 011 reports the module elaborating in ~24 s; my rebuild took 41 s for + the same module (33 s for CotMonotone vs 010's 27 s). Load-dependent, + not a reproducibility problem — both builds are green with identical + job counts (8658). +3. The conj-as-substitution convention is sound *because* every literal + in the definitions is rational (attack (i) below). This invariant is + stated nowhere in the .lean file itself (the record's mapping table + implies it). A future edit inserting a genuinely complex coefficient + would silently decouple the substitution from conjugation — worth a + one-line comment if the file is ever extended. Maintenance caveat, + not an error. + +## Claims that survive + +### (a) Statement mapping — CONFIRMED, object by object, no deltas + +Every definition in `LaurentBlock.lean` lined up against 005 §(ii)-A +Steps 2–4 and `plaw_general.closed_forms()` (read line by line): + +| Lean | 005 / plaw_general | verdict | +|---|---|---| +| `sinF z i = (z − z⁻¹)/(2i)`, `cosF z = (z + z⁻¹)/2` | `fsin`/`fcos` (Euler form, coefficient ±1/(2i), 1/2) | match | +| `B = sinF (x*y) i` | B = sin(α+β) = fsin(1,1,0,0) | match | +| `C = sinF y i * x` | C = sin β · e^{iα} | match | +| `A1 = B(1 − y⁻²)` | A₁ = R₀(A) = B(1−y⁻²) | match | +| `mu = y⁻²P⁻²Q²` | μ = e^{−2iaα+2i(b−1)β} = mono(0,−2,−2,2) | match | +| `delta = y⁻¹P⁻¹Q` | δ = e^{i(−aα+(b−1)β)}, δ² = μ | match | +| `w = B(1 − y⁻² + y⁻²P⁻² − μ)` | w = B(1 + P⁻²y⁻² − y⁻² − μ) | match (commuted) | +| `tau = w + mu·w[subst]` | τ = w + μ conj w | match | +| `proj y P Q i v vc = (delta y⁻¹P⁻¹Q⁻¹·v − delta·vc)/(2i)` | p(v) = Im(conj δ·v); note delta y⁻¹ P⁻¹ Q⁻¹ = yPQ⁻¹ = conj δ | match | +| `m = proj (w − τ/2) (…subst)/2` | m = Im(conj δ(w − τ/2))/2 | match | +| `v2 x y P i = B + y⁻²(P⁻²·C[subst] − B)` | v₂ = R₀(Rot_A(2aα)C) = B + y⁻²(P⁻² conj C − B); no Q anywhere (see (d)) | match | +| `reflAB p = (p.2, p.1)` | R₂ = conj | match | +| `reflCA x p = (x²p.2, x⁻²p.1)` | R₁ = e^{2iα} conj | match | +| `reflBC x y i p` | R₀ = B + e^{−2iβ} conj(·−B) | match | +| `rotA P p = (P²p.1, P⁻²p.2)` | Rot_A(2aα) about A = 0 | match (angle 2aα via P², factor-of-2 correct) | +| `rotB x y Q i p` | Rot_B(−2bβ) about B via Q⁻² | match (sign −2bβ correct) | + +Composition order checked against 005 Step 1's word-ordered product: +u = R₀ ∘ (R₁R₂)^a ∘ (R₀R₂)^b applies (R₀R₂)^b first — the Lean nesting +`reflBC (rotA P (rotB … p))` and the iterate nesting in +`half_word_letters` both match. `pair_collapse_A/B` are 005 Step 2 +exactly (R₁R₂ = Rot_A(2α) = `rotA` at P := x; R₀R₂ = Rot_B(−2β) = `rotB` +at Q := y). The RHS monomials of the identities decode to exactly 005's +angles: `cosF P` = cos(aα), `sinF Q i` = sin(bβ), `cosF (x*y*Q)` = +cos(α+(b+1)β), `sinF (y*Q) i` = sin((b+1)β), `sinF (P*Q⁻¹) i` = +sin(aα−bβ), `sinF (x⁻¹y⁻¹PQ⁻¹) i` = sin((a−1)α−(b+1)β) — each row also +float-verified at 200 random points (Layer 2, worst error 1.5e-15). +I1–I4/D1/D2 are 005's identities verbatim, same LHS orientation +(m − p(C) for I1; p(·) − m for I2–I4), same signs, no permutation. The +I1 = −(S+T), I4 = S−T structure is preserved. `glide_action` is 008 §1's +p∘u = 2m − p; `half_word_sq_is_translation` is Step 3's u² = z + τ with +conj(u(z)) expanded per the pair convention — checked to be the correct +expansion. The record's verbatim theorem block matches the file +character-for-character where quoted. + +**Attacks that found nothing:** hunted for a factor-of-2 slip in the +angle conventions (P² vs P in `rotA` — P² is right: Rot_A(2aα) with +P = e^{iaα}); for a sign flip in `rotB` (Q⁻² encodes −2bβ — right); for +an I2/I3 swap (cos(aα)·sin((b+1)β) vs cos((b+1)β)·sin(aα) — not +swapped); for an LHS orientation flip in I1 (it is m − p(C), the others +p(·) − m — matches 005); for a δ vs conj δ swap in `proj` (the first +factor is `delta y⁻¹ P⁻¹ Q⁻¹` = conj δ, correctly multiplying v, not +vc). Each of these would have made Layer 2's instance comparison against +005's prose formulas fail at 1e-15 precision; none did. + +### (b) The conjugation encoding — CONFIRMED; nothing smuggled + +- The substitution (x,y,P,Q,i) ↦ (x⁻¹,y⁻¹,P⁻¹,Q⁻¹,−i) equals complex + conjugation on the torus instance because (1) each generator has unit + modulus there, so inversion = conjugation, and (2) every literal + coefficient in the file is rational (attack (i): read all defs — the + constants are 1 and 2 only), and conjugating c(i) ∈ ℚ(i) is i ↦ −i. + This is exactly 008 §1's star involution ("x→1/x, y→1/y, i→−i is a + ring automorphism equal to complex conjugation on the torus"), + extended by P, Q inversion — which 008 had implicitly, since it + specialized P → x₀^a before starring. Same operation. +- Nothing in the block needs conj as an involution on *arbitrary* + elements: every conjugate the theorems consume is either the + substituted form of an explicit generator expression (I1–I4, D1, D2, + τ, m) or a second pair component that is *universally quantified* + (`glide_action` takes arbitrary z, zc; `half_word_closed_form` takes + an arbitrary pair p). That is the safe direction: the Lean theorems + are componentwise affine identities in the pair, so they are + *stronger* than the conjugate-pair reading — the intended semantics is + recovered by instantiating zc := conj z, with no hypothesis anywhere + demanding p.2 = conj p.1. Nothing is smuggled: the certificate never + assumes the second component is genuinely a conjugate; it proves + identities that hold for any second component, of which the conjugate + is one instance. +- The pair invariant itself (second components of the composite maps = + substituted first components) is proved by the theorems' RHS shapes, + and I verified it independently: symbolically in Layer 1 (the second + components of my recomposed maps match the substituted closed forms + exactly) and numerically in Layer 2 (`pair C`, `pair w`, `pair v2` + rows: substituted expression = complex conjugate at the instance, + ~2e-15). + +### (c) The dropped i² = −1 — SAFE; the homogeneity claim independently proven + +Direction check: the theorems are ∀-quantified over i ≠ 0, so the +intended instance i := Complex.I is an *instantiation* — the certificate +at the instance can only say more, never less, than 005's statement. +What remained to check is that at i = I the statements ARE 005's +statements: Layer 2 confirms sinF/cosF at the torus instance reduce to +the actual sines and cosines of exactly 005's angles (11 dictionary +rows, worst 1.5e-15), so nothing weaker is being said at the instance — +no identity's two sides differ by an i²-collapsible factor. Stronger: +Layer 1 re-proves every identity with i as a *formal Laurent variable* +in my own ring — an independent proof of 011's homogeneity finding +(each identity is an exact zero polynomial in ℚ[x^±,y^±,P^±,Q^±,i^±]; +had any side needed i² = −1, my formal-i check would have failed, since +my ring imposes no relation on i). 011's "discovered because ring_nf +never consumed i² = −1" is thus confirmed by a second, non-Lean method. + +### (d) Genericity — CONFIRMED; no silent narrowing + +- (a,b) enter the Tier A/B theorems only through the ∀-quantified + generators P, Q (grep: no natural-number parameter appears outside + `iterate_collapse_*`, `half_word_letters`, and the `Specialize` + section). One identity check covers all (a,b) — 005's move, faithfully. +- ℕ vs ℤ: 005 needs integer parameters a ≥ b ≥ 1. `half_word_letters` + (a b : ℕ) covers the genuine letter words for all a, b ≥ 1 (and adds + the degenerate a = 0 / b = 0 cases 005 never claims — extra scope in + the safe direction). The Tier C `*_spec` corollaries at a b : ℤ (zpow) + cover every integer pair, a superset of 005's range. The specialization + mechanism differs from 005's (instantiation of ∀-variables vs ring + homomorphism), as 011's table says; the conclusion is identical, and + the dangerous direction (formal zero ⟹ geometric zero) is inherited + from the ∀-statement — 008 §2's soundness argument is not even needed + in this encoding. +- v₂'s dropped Q: confirmed unused in 005 too. `plaw_general`'s v2 is + built from monomials with eQ = 0 throughout, and geometrically + v₂ = R₀(Rot_A(2aα)C) involves no b. I4's RHS still carries Q (through + p and m) — consistent in both. + +### (e) Cheat scan — CLEAN + +`grep -inE "sorry|admit|native_decide|decide|axiom|unsafe|partial|implemented_by|notation|macro|elab|opaque|extern"` +over all four project .lean files: only hits are the word "sorry" inside +two doc comments. No local notation, no instance declarations, no +shadowing (all defs live in the fresh `BilliardsFormal.Laurent` +namespace). Hypotheses x, y, P, Q, i ≠ 0 and (2 : K) ≠ 0 are jointly +satisfiable (ℚ, ℂ), satisfied at *every* torus instance (units are +nonzero; char 0), so nothing is vacuous and the char-2 exclusion costs +no intended instance. `BilliardsFormal.lean` imports both `CotMonotone` +and `LaurentBlock`, and my own build log shows all three modules +elaborated — the build cannot pass without elaborating the certificate. +`AxiomCheck.lean` is on disk (010's nit fixed) and enumerates all 23 + +009's 2 theorems; I did not rely on it (see (f)). + +### (f) Independent rebuild + axiom audit — GREEN + +From `problems/billiards-triangles/formal/`, PATH prefixed with +`$HOME/.elan/bin`: + +``` +$ rm -rf .lake/build # project output only; .lake/packages kept +$ lake build +✔ [8655/8658] Built BilliardsFormal.CotMonotone (33s) +✔ [8656/8658] Built BilliardsFormal.LaurentBlock (41s) +✔ [8657/8658] Built BilliardsFormal (7.1s) +Build completed successfully (8658 jobs). +``` + +Then my own scratch `SkepticAxiomCheck011.lean` (`import BilliardsFormal` ++ `#print axioms` on each of the 23 `Laurent` theorems, the list +enumerated by me from the source): all 23 report exactly +`[propext, Classical.choice, Quot.sound]` — no `sorryAx`, no +`Lean.ofReduceBool` (so no `native_decide` anywhere in the dependency +cone), no project-local axiom. Scratch file deleted after the run. +Independence scope as in 010: fresh session, different agent, project +build products deleted and re-elaborated from committed source; mathlib +oleans from the pinned on-disk cache (the trust base every mathlib user +accepts). Toolchain pins consistent across `lean-toolchain`, +`lakefile.toml`, `lake-manifest.json` (mathlib rev +`905b95818eb32af7874a58b427f50c1711a5e96c` = tag v4.32.2), and both +build logs (attack (j)). + +### (g) Record audit — HONEST + +- The bridge exclusion is genuine and prominent: 011's "Not attacked, + deliberately" paragraph, the module docstring's "What is NOT proved + (the bridge)" section, and the index entry's + `geometry-bridge-not-formalized` gap all scope it out identically, and + nothing in the record's Outcome-equivalent sections claims geometric + content for the certificate. +- The numeric i = 0.7 − 1.3j check is labelled a pre-proof sanity check + ("the Lean proofs supersede it") — correctly not load-bearing. +- prior-art.json 011 entry: `range` names exactly the formal statements + and the encoding; `one_line` matches the record (23 theorems, tiers + A+B+C, i-homogeneity, bridge out of scope, review pending); `gaps` + listed the two pending pass parts plus the permanent bridge gap — + the correct under-claiming direction. The two pending gaps are closed + by this record (index updated; 011's .md untouched); the bridge gap + stays, as permanent scope. +- Numbers spot-checked: 23 theorems (counted in source: 6 + 11 + 6), + 8658 jobs (my rebuild agrees), zero sorries (kernel + grep agree), + "~420 lines" vs 423 (nit 1). + +### (h) Spot re-derivation — ALL identities re-proven from scratch; D1 also by hand + +- **Layer 1** (`lbsk_review.py`, `data/lbsk_review.json`): my own + 7-variable Laurent ring over ℚ (exponent-tuple dict, Fraction + coefficients, division only by unit monomials — written for this + review, no code or design shared with `plaw_general.py`, which + represents coefficients in ℚ(i) with i NON-formal, or with 008's + `psk_review.py`). The Lean definitions were transcribed from the .lean + file and every theorem re-checked by exact polynomial subtraction: + **31/31 exact zero polynomials** — I1–I4, D1, D2, μ = δ², τ ∥ axis, + μ·conj μ = 1, u² = z + τ (z formal), glide action (z, zc formal), + both components of the closed form (pair formal), both pair collapses, + and `half_word_letters` re-composed by genuine symbolic iteration for + all 0 ≤ a, b ≤ 3 (16 cases, both components). This includes at least + one of I1–I4 as required — in fact all four. +- **By hand** (computer-free second witness), D1: p(B) − p(C) = + Im(conj δ·B) − Im(conj δ·C) with conj δ = e^{i(aα−(b−1)β)} gives + sin(α+β) sin(aα−(b−1)β) − sin β sin((a+1)α−(b−1)β); product-to-sum on + both, the (b−2)-terms cancel, leaving + ½[cos((a−1)α−bβ) − cos((a+1)α−bβ)] = sin α · sin(aα−bβ). This is + 005's D1 and Lean's `sinF x i * sinF (P*Q⁻¹) i` exactly. +- **Layer 2** closes the mapping triangle: Lean statement (proven true + by Layer 1 and the kernel) ↔ 005's prose formulas (I1–I4/D1/D2 trig + RHS, Step-2 reflections, Step-3 closed forms μ, δ, w, τ, m, the + p(v) = Im(conj δ v) projection, v₂ = R₀(Rot_A C), and u composed + directly from 005's geometric maps) — 30 mapped quantities, 200 random + (a, b, α, β) with a ≤ 12, worst error 4.4e-15. + +### Verdict on the three-part FORMALIZED pass (docs/FORMALIZE.md) + +1. **Statement review** (the point): DONE by this record, by a + non-formalizer, definition by definition and theorem by theorem — no + narrowing, no deltas; the encoding table of 011 is signed off + unchanged. +2. **Clean build**: CONFIRMED — zero sorries, axioms exactly the + standard three for all 23 theorems, full source in the repo, verified + by my own audit from my own theorem list. +3. **Recorded toolchain**: CONFIRMED — pins consistent, and my own + independent `lake build` (fresh session, build output deleted) is + green. + +**The FORMALIZED status of 011 (= the machine-checked certificate for +005's Laurent-identity block) is complete.** Scope unchanged from 011's +own statement: the ring-model block only. Not claimed by this review: +the geometry bridge (permanently out of the certificate's scope — 005's +Steps 1–3 prose + 008's off-torus recomposition carry it, and Layer 2's +float agreement with 005's geometric maps is a mapping check at 200 +points, not a proof); Lemmas C/D, the case tree, and the sufficiency +segments (VERIFIED via 008, not formal); mathlib internals; the Lean +kernel. + +## Leads generated + +Nothing new beyond 011's own leads, which this review leaves as the +queue state (Lemma D / C-chain formalization; the torus-evaluation +lemma `sinF (exp (θI)) I = sin θ`, which would move Layer 2's float +mapping rows into the kernel; the i-homogeneity Python check of 005's +coefficient dictionaries — note Layer 1 here already proves the +homogeneity *of the identity block*; 011's lead 2 asks the finer +per-coefficient question about `plaw_general`'s ring). One addition: + +1. **Comment the rationality invariant** (nit 3): a one-line note in + `LaurentBlock.lean` that the conj-as-substitution convention requires + all literal coefficients rational. Definite outcome: the invariant is + stated where a future editor will see it, or deliberately left to the + record. + +## References + +- `problems/billiards-triangles/attempts/011-lean-laurent-block.md` (the + record under review). +- `problems/billiards-triangles/attempts/005-complete-death-law-theorem.md` + §(ii)-A Steps 1–4 (the statements of truth), §(i) Reduction (glide + action use site), Conventions block. +- `problems/billiards-triangles/attempts/008-skeptic-review-of-005.md` + §§1–2 (star involution, off-torus pair algebra, v₂ identification — + the second witness). +- `problems/billiards-triangles/attempts/009-…`/`010-…` (lane precedent; + pass shape). +- `docs/FORMALIZE.md` (the three-part pass this record completes). +- `problems/billiards-triangles/explore/plaw_general.py` + (`closed_forms`, `cmd_formal` — read line by line; never imported by + my tool). +- `problems/billiards-triangles/formal/` — `BilliardsFormal/LaurentBlock.lean` + (every line), `BilliardsFormal.lean`, `AxiomCheck.lean`, + `lakefile.toml`, `lean-toolchain`, `lake-manifest.json`, + `lake-build.log`. +- New code and data: `explore/lbsk_review.py`, `data/lbsk_review.json`. +- mathlib4 tag v4.32.2 (rev `905b95818eb32af7874a58b427f50c1711a5e96c`), + pinned checkout. No papers; no machine transcriptions. diff --git a/problems/billiards-triangles/data/lbsk_review.json b/problems/billiards-triangles/data/lbsk_review.json new file mode 100644 index 0000000..0a1186a --- /dev/null +++ b/problems/billiards-triangles/data/lbsk_review.json @@ -0,0 +1,72 @@ +{ + "symbolic_checks": { + "I1": true, + "I2": true, + "I3": true, + "I4": true, + "D1": true, + "D2": true, + "mu_eq_delta_sq": true, + "tau_parallel_axis": true, + "mu_mul_conj_mu": true, + "half_word_sq": true, + "glide_action": true, + "half_word_closed_form.1": true, + "half_word_closed_form.2": true, + "pair_collapse_A": true, + "pair_collapse_B": true, + "half_word_letters[0,0]": true, + "half_word_letters[0,1]": true, + "half_word_letters[0,2]": true, + "half_word_letters[0,3]": true, + "half_word_letters[1,0]": true, + "half_word_letters[1,1]": true, + "half_word_letters[1,2]": true, + "half_word_letters[1,3]": true, + "half_word_letters[2,0]": true, + "half_word_letters[2,1]": true, + "half_word_letters[2,2]": true, + "half_word_letters[2,3]": true, + "half_word_letters[3,0]": true, + "half_word_letters[3,1]": true, + "half_word_letters[3,2]": true, + "half_word_letters[3,3]": true + }, + "n_symbolic": 31, + "n_symbolic_pass": 31, + "float_worst_errors": { + "cosF_P=cos(a.al)": 1.2412670766236366e-16, + "sinF_P=sin(a.al)": 1.2412670766236366e-16, + "sinF_Q=sin(b.be)": 1.2412670766236366e-16, + "sinF_x=sin(al)": 1.5700924586837752e-16, + "sinF_y=sin(be)": 1.2412670766236366e-16, + "sinF_xy=sin(al+be)": 3.3766115072321297e-16, + "cosF_xyQ=cos(al+(b+1)be)": 9.226380847435024e-16, + "sinF_yQ=sin((b+1)be)": 4.443059973708341e-16, + "cosF_yQ=cos((b+1)be)": 1.5275656190879576e-15, + "sinF_PQinv=sin(a.al-b.be)": 8.345155887641791e-16, + "sinF_D2mono=sin((a-1)al-(b+1)be)": 1.4443570596498132e-15, + "B=sin(al+be)": 3.3766115072321297e-16, + "C=sin(be)e^{i.al}": 1.5700924586837752e-16, + "mu=e^{-2ia.al+2i(b-1)be}": 2.278663452567752e-15, + "delta": 1.1801832636420706e-15, + "w(005 Step3)": 3.1106063782808446e-15, + "tau": 4.445140139234677e-15, + "m=Im(conj.delta(w-tau/2))/2": 1.3394761424494089e-15, + "proj=Im(conj.delta.v)": 6.661338147750939e-16, + "v2=R0(RotA C)": 9.305364597889227e-16, + "u=mu.conj z+w": 1.4043333874306805e-15, + "pair C": 3.5135752366389115e-16, + "pair w": 1.9860273225978185e-15, + "pair v2": 9.42055475210265e-16, + "I1 trig": 6.674335893081556e-16, + "I2 trig": 5.578801654593729e-16, + "I3 trig": 1.3340734093775516e-15, + "I4 trig": 7.217783939387125e-16, + "D1 trig": 7.406111315403549e-16, + "D2 trig": 1.276831898851135e-15 + }, + "float_overall_worst": 4.445140139234677e-15, + "ok": true, + "meaning": "Layer 1: every Lean theorem statement of 011 re-proven as an exact zero polynomial in an independent 7-variable Laurent ring over Q with i FORMAL; Layer 2: the Lean statements at the torus instance match 005's prose quantities to float precision" +} \ No newline at end of file diff --git a/problems/billiards-triangles/explore/lbsk_review.py b/problems/billiards-triangles/explore/lbsk_review.py new file mode 100644 index 0000000..df9862a --- /dev/null +++ b/problems/billiards-triangles/explore/lbsk_review.py @@ -0,0 +1,387 @@ +#!/usr/bin/env python3 +"""Attempt 012: skeptic pass on attempt 011 (Lean LaurentBlock.lean) -- +independent re-derivation, written from the LEAN FILE's definitions (not +from plaw_general.py, whose ring this script does not import). + +Layer 1 (SYMBOLIC, exact): a from-scratch Laurent-polynomial ring over Q in +SEVEN variables (x, y, P, Q, i, z1, z2) -- i is a formal unit exactly as in +the Lean statements (no i^2 = -1). Every division in the Lean definitions +is by a unit monomial (2, 2i, generator inverses), so the whole block lives +inside the Laurent ring and each identity is checked by exact polynomial +subtraction. This re-proves the Lean statements independently of both the +Lean kernel and 005's/plaw_general's code. + +Layer 2 (FLOAT, mapping): instantiate x=e^{ia_}, y=e^{ib_}, P=e^{i a a_}, +Q=e^{i b b_}, i=1j and check the Lean-side expressions against 005's OWN +prose formulas (trig right-hand sides of I1-I4/D1/D2, Step-2 reflection +formulas, Step-3 closed forms mu/delta/w/tau/m, p(v)=Im(conj delta v)). +This nails the statement mapping: the Lean theorems at the intended +instance say exactly what 005 says. + +stdlib only; deterministic (fixed seed). Usage: + python3 lbsk_review.py [--out data/lbsk_review.json] +""" +import argparse +import cmath +import json +import math +import random +import sys +from fractions import Fraction as Fr + +# ---------------------------------------------------------------- Layer 1 +# poly: dict {(ex,ey,eP,eQ,ei,ez1,ez2): Fraction}, Laurent exponents. + +NV = 7 +X = 0; Y = 1; PP = 2; QQ = 3; II = 4; Z1 = 5; Z2 = 6 + +def mono(coef=Fr(1), **exps): + e = [0]*NV + for k, v in exps.items(): + e[{'x':X,'y':Y,'P':PP,'Q':QQ,'i':II,'z1':Z1,'z2':Z2}[k]] = v + return {tuple(e): Fr(coef)} + +ONE = mono() +def const(c): return mono(Fr(c)) + +def padd(p, q): + r = dict(p) + for k, c in q.items(): + r[k] = r.get(k, Fr(0)) + c + if r[k] == 0: del r[k] + return r + +def pneg(p): return {k: -c for k, c in p.items()} +def psub(p, q): return padd(p, pneg(q)) + +def pmul(p, q): + r = {} + for k1, c1 in p.items(): + for k2, c2 in q.items(): + k = tuple(a+b for a, b in zip(k1, k2)) + r[k] = r.get(k, Fr(0)) + c1*c2 + if r[k] == 0: del r[k] + return r + +def is_mono(p): return len(p) == 1 + +def pinv(p): + """Inverse of a unit monomial.""" + assert is_mono(p), "inverting a non-monomial: encoding broken" + (k, c), = p.items() + return {tuple(-e for e in k): Fr(1)/c} + +def pdiv_unit(p, u): # divide by unit monomial u + return pmul(p, pinv(u)) + +def ppow(p, n): + r = ONE + for _ in range(n): r = pmul(r, p) + return r + +def zero(p): return p == {} + +# generators +gx, gy, gP, gQ, gi = (mono(x=1), mono(y=1), mono(P=1), mono(Q=1), mono(i=1)) +gz1, gz2 = mono(z1=1), mono(z2=1) +TWO = const(2) + +# --- Lean definitions, transcribed line by line from LaurentBlock.lean --- +def sinF(z, i): # (z - z^-1)/(2*i) + return pdiv_unit(psub(z, pinv(z)), pmul(TWO, i)) + +def cosF(z): # (z + z^-1)/2 + return pdiv_unit(padd(z, pinv(z)), TWO) + +def Bd(x, y, i): return sinF(pmul(x, y), i) +def Cd(x, y, i): return pmul(sinF(y, i), x) +def A1d(x, y, i): return pmul(Bd(x, y, i), psub(ONE, ppow(pinv(y), 2))) +def mud(y, P, Q): + return pmul(pmul(ppow(pinv(y), 2), ppow(pinv(P), 2)), ppow(Q, 2)) +def deltad(y, P, Q): return pmul(pmul(pinv(y), pinv(P)), Q) +def wd(x, y, P, Q, i): + inner = psub(padd(psub(ONE, ppow(pinv(y), 2)), + pmul(ppow(pinv(y), 2), ppow(pinv(P), 2))), + mud(y, P, Q)) + return pmul(Bd(x, y, i), inner) +def taud(x, y, P, Q, i): + return padd(wd(x, y, P, Q, i), + pmul(mud(y, P, Q), + wd(pinv(x), pinv(y), pinv(P), pinv(Q), pneg(i)))) +def projd(y, P, Q, i, v, vc): + num = psub(pmul(deltad(pinv(y), pinv(P), pinv(Q)), v), + pmul(deltad(y, P, Q), vc)) + return pdiv_unit(num, pmul(TWO, i)) +def md(x, y, P, Q, i): + wc = wd(pinv(x), pinv(y), pinv(P), pinv(Q), pneg(i)) + tc = taud(pinv(x), pinv(y), pinv(P), pinv(Q), pneg(i)) + t = taud(x, y, P, Q, i) + v = psub(wd(x, y, P, Q, i), pdiv_unit(t, TWO)) + vc = psub(wc, pdiv_unit(tc, TWO)) + return pdiv_unit(projd(y, P, Q, i, v, vc), TWO) +def v2d(x, y, P, i): + B_ = Bd(x, y, i) + return padd(B_, pmul(ppow(pinv(y), 2), + psub(pmul(ppow(pinv(P), 2), + Cd(pinv(x), pinv(y), pneg(i))), B_))) + +# maps on pairs +def reflAB(p): return (p[1], p[0]) +def reflCA(x, p): return (pmul(ppow(x, 2), p[1]), pmul(ppow(pinv(x), 2), p[0])) +def reflBC(x, y, i, p): + B_ = Bd(x, y, i); Bc = Bd(pinv(x), pinv(y), pneg(i)) + return (padd(B_, pmul(ppow(pinv(y), 2), psub(p[1], B_))), + padd(Bc, pmul(ppow(pinv(pinv(y)), 2), psub(p[0], Bc)))) +def rotA(P, p): return (pmul(ppow(P, 2), p[0]), pmul(ppow(pinv(P), 2), p[1])) +def rotB(x, y, Q, i, p): + B_ = Bd(x, y, i); Bc = Bd(pinv(x), pinv(y), pneg(i)) + return (padd(B_, pmul(ppow(pinv(Q), 2), psub(p[0], B_))), + padd(Bc, pmul(ppow(pinv(pinv(Q)), 2), psub(p[1], Bc)))) + +def run_symbolic(): + x, y, P, Q, i = gx, gy, gP, gQ, gi + xc, yc, Pc, Qc, ic = pinv(x), pinv(y), pinv(P), pinv(Q), pneg(i) + checks = {} + m_ = md(x, y, P, Q, i) + pr = lambda v, vc: projd(y, P, Q, i, v, vc) + # I1 + lhs = psub(m_, pr(Cd(x, y, i), Cd(xc, yc, ic))) + rhs = pneg(padd(pmul(pmul(cosF(P), sinF(x, i)), sinF(Q, i)), + pmul(pmul(sinF(P, i), sinF(y, i)), + cosF(pmul(pmul(x, y), Q))))) + checks['I1'] = zero(psub(lhs, rhs)) + # I2 + lhs = psub(pr(A1d(x, y, i), A1d(xc, yc, ic)), m_) + rhs = pmul(pmul(cosF(P), sinF(pmul(y, Q), i)), sinF(pmul(x, y), i)) + checks['I2'] = zero(psub(lhs, rhs)) + # I3 + lhs = psub(pr(Bd(x, y, i), Bd(xc, yc, ic)), m_) + rhs = pmul(pmul(cosF(pmul(y, Q)), sinF(P, i)), sinF(pmul(x, y), i)) + checks['I3'] = zero(psub(lhs, rhs)) + # I4 + lhs = psub(pr(v2d(x, y, P, i), v2d(xc, yc, Pc, ic)), m_) + rhs = psub(pmul(pmul(cosF(P), sinF(x, i)), sinF(Q, i)), + pmul(pmul(sinF(P, i), sinF(y, i)), + cosF(pmul(pmul(x, y), Q)))) + checks['I4'] = zero(psub(lhs, rhs)) + # D1 + lhs = psub(pr(Bd(x, y, i), Bd(xc, yc, ic)), + pr(Cd(x, y, i), Cd(xc, yc, ic))) + rhs = pmul(sinF(x, i), sinF(pmul(P, pinv(Q)), i)) + checks['D1'] = zero(psub(lhs, rhs)) + # D2 + lhs = psub(pr(A1d(x, y, i), A1d(xc, yc, ic)), + pr(Cd(x, y, i), Cd(xc, yc, ic))) + rhs = pneg(pmul(sinF(y, i), + sinF(pmul(pmul(pmul(pinv(x), pinv(y)), P), pinv(Q)), i))) + checks['D2'] = zero(psub(lhs, rhs)) + # glide facts + checks['mu_eq_delta_sq'] = zero(psub(mud(y, P, Q), + ppow(deltad(y, P, Q), 2))) + checks['tau_parallel_axis'] = zero( + pr(taud(x, y, P, Q, i), taud(xc, yc, Pc, Qc, ic))) + checks['mu_mul_conj_mu'] = zero(psub( + pmul(mud(y, P, Q), mud(yc, Pc, Qc)), ONE)) + # half_word_sq_is_translation (z1 formal) + wc = wd(xc, yc, Pc, Qc, ic) + lhs = padd(pmul(mud(y, P, Q), padd(pmul(mud(yc, Pc, Qc), gz1), wc)), + wd(x, y, P, Q, i)) + checks['half_word_sq'] = zero(psub(lhs, padd(gz1, taud(x, y, P, Q, i)))) + # glide_action (z1, z2 formal) + lhs = pr(padd(pmul(mud(y, P, Q), gz2), wd(x, y, P, Q, i)), + padd(pmul(mud(yc, Pc, Qc), gz1), wc)) + rhs = psub(pmul(TWO, m_), pr(gz1, gz2)) + checks['glide_action'] = zero(psub(lhs, rhs)) + # half_word_closed_form (pair (z1,z2) formal) + got = reflBC(x, y, i, rotA(P, rotB(x, y, Q, i, (gz1, gz2)))) + want = (padd(pmul(mud(y, P, Q), gz2), wd(x, y, P, Q, i)), + padd(pmul(mud(yc, Pc, Qc), gz1), wc)) + checks['half_word_closed_form.1'] = zero(psub(got[0], want[0])) + checks['half_word_closed_form.2'] = zero(psub(got[1], want[1])) + # pair collapses + got = reflCA(x, reflAB((gz1, gz2))) + want = rotA(x, (gz1, gz2)) + checks['pair_collapse_A'] = (zero(psub(got[0], want[0])) + and zero(psub(got[1], want[1]))) + got = reflBC(x, y, i, reflAB((gz1, gz2))) + want = rotB(x, y, y, i, (gz1, gz2)) + checks['pair_collapse_B'] = (zero(psub(got[0], want[0])) + and zero(psub(got[1], want[1]))) + # half_word_letters for a,b in 0..3 (iterates computed symbolically) + for a in range(4): + for b in range(4): + p = (gz1, gz2) + for _ in range(b): + p = reflBC(x, y, i, reflAB(p)) + for _ in range(a): + p = reflCA(x, reflAB(p)) + p = reflBC(x, y, i, p) + Pa, Qb = ppow(x, a), ppow(y, b) + Pac, Qbc = pinv(Pa), pinv(Qb) + want = (padd(pmul(mud(y, Pa, Qb), gz2), wd(x, y, Pa, Qb, i)), + padd(pmul(mud(yc, Pac, Qbc), gz1), + wd(xc, yc, Pac, Qbc, ic))) + checks[f'half_word_letters[{a},{b}]'] = ( + zero(psub(p[0], want[0])) and zero(psub(p[1], want[1]))) + return checks + +# ---------------------------------------------------------------- Layer 2 +def crun(f, *args): return f(*args) + +def run_float(trials=200, seed=20260804): + rng = random.Random(seed) + worst = {} + def upd(tag, err): + worst[tag] = max(worst.get(tag, 0.0), err) + + # numeric versions of the Lean defs (complex arithmetic, i passed in) + def nsinF(z, i): return (z - 1/z) / (2*i) + def ncosF(z): return (z + 1/z) / 2 + def nB(x, y, i): return nsinF(x*y, i) + def nC(x, y, i): return nsinF(y, i) * x + def nA1(x, y, i): return nB(x, y, i) * (1 - (1/y)**2) + def nmu(y, P, Q): return (1/y)**2 * (1/P)**2 * Q**2 + def ndelta(y, P, Q): return (1/y) * (1/P) * Q + def nw(x, y, P, Q, i): + return nB(x, y, i) * (1 - y**-2 + y**-2 * P**-2 - nmu(y, P, Q)) + def ntau(x, y, P, Q, i): + return (nw(x, y, P, Q, i) + + nmu(y, P, Q) * nw(1/x, 1/y, 1/P, 1/Q, -i)) + def nproj(y, P, Q, i, v, vc): + return (ndelta(1/y, 1/P, 1/Q)*v - ndelta(y, P, Q)*vc) / (2*i) + def nm(x, y, P, Q, i): + wc = nw(1/x, 1/y, 1/P, 1/Q, -i) + tc = ntau(1/x, 1/y, 1/P, 1/Q, -i) + return nproj(y, P, Q, i, + nw(x, y, P, Q, i) - ntau(x, y, P, Q, i)/2, + wc - tc/2) / 2 + def nv2(x, y, P, i): + return (nB(x, y, i) + + y**-2 * (P**-2 * nC(1/x, 1/y, -i) - nB(x, y, i))) + + for _ in range(trials): + a = rng.randint(1, 12) + b = rng.randint(1, a) + al = rng.uniform(0.05, 1.3) + be = rng.uniform(0.05, min(1.3, math.pi - 0.1 - al)) + x = cmath.exp(1j*al); y = cmath.exp(1j*be) + P = cmath.exp(1j*a*al); Q = cmath.exp(1j*b*be) + i = 1j + Bg = math.sin(al+be) # 005: B = sin(alpha+beta), real + # -- the trig dictionary of the Lean statements vs 005's prose + upd('cosF_P=cos(a.al)', abs(ncosF(P) - math.cos(a*al))) + upd('sinF_P=sin(a.al)', abs(nsinF(P, i) - math.sin(a*al))) + upd('sinF_Q=sin(b.be)', abs(nsinF(Q, i) - math.sin(b*be))) + upd('sinF_x=sin(al)', abs(nsinF(x, i) - math.sin(al))) + upd('sinF_y=sin(be)', abs(nsinF(y, i) - math.sin(be))) + upd('sinF_xy=sin(al+be)', abs(nsinF(x*y, i) - math.sin(al+be))) + upd('cosF_xyQ=cos(al+(b+1)be)', + abs(ncosF(x*y*Q) - math.cos(al+(b+1)*be))) + upd('sinF_yQ=sin((b+1)be)', abs(nsinF(y*Q, i) - math.sin((b+1)*be))) + upd('cosF_yQ=cos((b+1)be)', abs(ncosF(y*Q) - math.cos((b+1)*be))) + upd('sinF_PQinv=sin(a.al-b.be)', + abs(nsinF(P/Q, i) - math.sin(a*al-b*be))) + upd('sinF_D2mono=sin((a-1)al-(b+1)be)', + abs(nsinF(P/(x*y*Q), i) - math.sin((a-1)*al-(b+1)*be))) + # -- 005 Step 2/3 objects vs Lean defs at the instance + upd('B=sin(al+be)', abs(nB(x, y, i) - Bg)) + upd('C=sin(be)e^{i.al}', abs(nC(x, y, i) + - math.sin(be)*cmath.exp(1j*al))) + mu5 = cmath.exp(-2j*a*al + 2j*(b-1)*be) + upd('mu=e^{-2ia.al+2i(b-1)be}', abs(nmu(y, P, Q) - mu5)) + de5 = cmath.exp(1j*(-a*al + (b-1)*be)) + upd('delta', abs(ndelta(y, P, Q) - de5)) + w5 = Bg * (1 + cmath.exp(-2j*(a*al+be)) - cmath.exp(-2j*be) - mu5) + upd('w(005 Step3)', abs(nw(x, y, P, Q, i) - w5)) + tau5 = w5 + mu5 * w5.conjugate() + upd('tau', abs(ntau(x, y, P, Q, i) - tau5)) + m5 = ((de5.conjugate() * (w5 - tau5/2)).imag) / 2 + upd('m=Im(conj.delta(w-tau/2))/2', abs(nm(x, y, P, Q, i) - m5)) + # p(v) = Im(conj delta . v) at a random point, pair = (v, conj v) + v = complex(rng.uniform(-1, 1), rng.uniform(-1, 1)) + upd('proj=Im(conj.delta.v)', + abs(nproj(y, P, Q, i, v, v.conjugate()) + - (de5.conjugate()*v).imag)) + # v2 = R0(RotA(2a.al) C) computed from 005's geometric maps + R0 = lambda z: Bg + cmath.exp(-2j*be) * (z - Bg).conjugate() + v2g = R0(cmath.exp(2j*a*al) * math.sin(be)*cmath.exp(1j*al)) + upd('v2=R0(RotA C)', abs(nv2(x, y, P, i) - v2g)) + # closed form u(z) vs direct composition of 005's maps + z = complex(rng.uniform(-1, 1), rng.uniform(-1, 1)) + zz = Bg + cmath.exp(-2j*b*be) * (z - Bg) # Rot_B(-2b.be) + zz = cmath.exp(2j*a*al) * zz # Rot_A(2a.al) + zz = R0(zz) # R_0 + upd('u=mu.conj z+w', abs(nmu(y, P, Q)*z.conjugate() + + nw(x, y, P, Q, i) - zz)) + # pair second components genuinely conjugate at the instance + upd('pair C', abs(nC(1/x, 1/y, -i) - nC(x, y, i).conjugate())) + upd('pair w', abs(nw(1/x, 1/y, 1/P, 1/Q, -i) + - nw(x, y, P, Q, i).conjugate())) + upd('pair v2', abs(nv2(1/x, 1/y, 1/P, -i) + - nv2(x, y, P, i).conjugate())) + # -- 005's I1-I4/D1/D2 trig RHS vs the Lean-statement LHS + m_ = nm(x, y, P, Q, i) + pC = nproj(y, P, Q, i, nC(x, y, i), nC(1/x, 1/y, -i)) + pB = nproj(y, P, Q, i, nB(x, y, i), nB(1/x, 1/y, -i)) + pA1 = nproj(y, P, Q, i, nA1(x, y, i), nA1(1/x, 1/y, -i)) + pv2 = nproj(y, P, Q, i, nv2(x, y, P, i), nv2(1/x, 1/y, 1/P, -i)) + S = math.cos(a*al)*math.sin(al)*math.sin(b*be) + T = math.sin(a*al)*math.sin(be)*math.cos(al+(b+1)*be) + upd('I1 trig', abs((m_ - pC) - (-(S + T)))) + upd('I2 trig', abs((pA1 - m_) + - math.cos(a*al)*math.sin((b+1)*be) + * math.sin(al+be))) + upd('I3 trig', abs((pB - m_) + - math.cos((b+1)*be)*math.sin(a*al) + * math.sin(al+be))) + upd('I4 trig', abs((pv2 - m_) - (S - T))) + upd('D1 trig', abs((pB - pC) + - math.sin(al)*math.sin(a*al-b*be))) + upd('D2 trig', abs((pA1 - pC) + - (-math.sin(be) + * math.sin((a-1)*al-(b+1)*be)))) + return worst + +def main(): + ap = argparse.ArgumentParser() + ap.add_argument("--out") + args = ap.parse_args() + print("=== Layer 1: symbolic (exact, 7-var Laurent ring over Q) ===") + checks = run_symbolic() + ok = True + for k, v in checks.items(): + if not v: + print(f" FAIL {k}") + ok = False + n = len(checks) + print(f" {sum(checks.values())}/{n} identities are exact zero " + f"polynomials (i FORMAL, no i^2=-1)") + print("=== Layer 2: float mapping to 005's prose (200 random " + "(a,b,al,be), a<=12) ===") + worst = run_float() + w = max(worst.values()) + for k in sorted(worst, key=worst.get, reverse=True)[:5]: + print(f" worst[{k}] = {worst[k]:.3e}") + print(f" overall worst error = {w:.3e} over {len(worst)} mapped " + f"quantities") + ok = ok and w < 1e-9 + print("SKEPTIC-011 SYMBOLIC+MAPPING:", "ALL OK" if ok else "FAILURES") + if args.out: + json.dump({"symbolic_checks": checks, + "n_symbolic": n, + "n_symbolic_pass": sum(checks.values()), + "float_worst_errors": worst, + "float_overall_worst": w, + "ok": ok, + "meaning": ("Layer 1: every Lean theorem statement of " + "011 re-proven as an exact zero polynomial " + "in an independent 7-variable Laurent ring " + "over Q with i FORMAL; Layer 2: the Lean " + "statements at the torus instance match " + "005's prose quantities to float " + "precision")}, + open(args.out, "w"), indent=1) + sys.exit(0 if ok else 1) + +if __name__ == "__main__": + main() diff --git a/problems/billiards-triangles/prior-art.json b/problems/billiards-triangles/prior-art.json index 9c97a8a..da4dd37 100644 --- a/problems/billiards-triangles/prior-art.json +++ b/problems/billiards-triangles/prior-art.json @@ -1,320 +1,342 @@ { - "problem": "billiards-triangles", - "route_status": "EVIDENCE", - "route_summary": "Word-census route mapped blind to length 26: 376 exact certificates (family orbits to length 66, gamma <= 160 deg), coverage staircase stalling at 135 deg, and a measured hyperbolic (not super-exponential) growth of needed word length vs angle - but certified box width collapses exponentially in length, so length buys angle reach, not area. Negatives are sample-bounded; skeptic-confirmed with one numeric correction.", - "attempts": [ - { - "id": "001", - "file": "attempts/001-word-census-coverage-map.md", - "date": "2026-07-30", - "mode": "blind", - "status": "EVIDENCE", - "mechanism": [ - "unfolding-word-census", - "certified-box-coverage", - "exhaustive-enumeration" - ], - "one_line": "Canonical word census to length 26 with 376 exact certificates: coverage staircase stalls at 135 deg, blind rediscovery of the ~112.5 deg frontier, and a hyperbolic certificate-length growth law along the W(a,b) family (checked to length 66).", - "leak_terms": [ - "112.4989", - "W(a,b)", - "coverage staircase", - "length buys angle reach", - "17,527" - ], - "range": "canonical translation words of even length <= 26 (17,527 words); 1/64-cell midpoint screen of the obtuse half-disk; exact dyadic-box anchors; family rows to W(8,8), gamma <= 160 deg", - "gaps": [ - "sample-bounded-negatives", - "death-angle-laws-unproven" - ] - }, - { - "id": "002", - "file": "attempts/002-skeptic-review-of-001.md", - "date": "2026-07-30", - "mode": "blind", - "status": "VERIFIED_WITH_CORRECTIONS", - "mechanism": [ - "adversarial-verification", - "independent-reimplementation", - "unfolding-word-census" - ], - "one_line": "From-scratch re-implementation of every layer confirms 001 (36/36 sampled certificates at 180 exact apexes, 17,527-word universe reproduced word-for-word, death law survives two out-of-sample prediction rows); certified-area figure corrected to ~4.3e-6.", - "leak_terms": [ - "147.5974", - "157.4976", - "4.3e-6" - ], - "verifies": "001", - "corrections": [ - "certified-area total over all 376 anchors is ~4.3e-6, not ~2.2e-6 (which is the map-anchor subtotal)", - "the 64.3% uncovered figure is of the resolved interior, not the whole obtuse region", - "anchor simulation re-derivation is at the anchor centre; the full 9-sample box attack ran on every 5th/7th record", - "kill-condition verdict is measured only to gamma ~ 160 deg; beyond that the hyperbolic form is the SPECULATION law's extrapolation" - ] - }, - { - "id": "003", - "file": "attempts/003-death-angle-laws.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED", - "mechanism": [ - "symbolic-trig-closed-form", - "certified-interval-inequality", - "out-of-sample-prediction", - "unfolding-word-census" - ], - "one_line": "The W(a,b) death-angle laws of 001 survive both out-of-sample tests (W(10,9) -> 162.90, W(12,12) -> 2160/13, to ~1e-12), and are then derived: exact division-free unfolding gives closed-form binding functions, a general law gamma_d = 180 - 90(a+b)/(a(b+1)) (confirmed out-of-sample on W(5,3) -> 144 and W(4,2) -> 135), and a machine-certified necessity theorem: the corridor is empty at every gamma >= gamma_d, for 15 verified members.", - "leak_terms": [ - "gamma_d = 180 - 90(a+b)/(a(b+1))", - "90(a+b)/(a(b+1))", - "glide reflection reduction", - "W(5,3)", - "144.0000000000", - "deathlaw_", - "Lemma C", - "division-free unfolding" - ], - "range": "necessity theorem machine-certified for the 15 members (2,1),(2,2),(3,2),(3,3),(4,2),(4,3),(4,4),(5,3),(5,4),(5,5),(6,6),(7,7),(8,8),(10,9),(12,12); exact alive certificates at gamma_d - 1e-6 (1e-8 for (3,3),(4,2)) for 8 members; float deaths measured to ~1e-11 for 11 members; parametric statement for all a >= b >= 1, a >= 2, a <= 2b+3 rests on two per-member-verified inputs (identities I1-I3 and Lemma C) labelled SPECULATION in general", - "gaps": [ - "sufficiency-limit-unclosed (death = gamma_d exactly needs alive points approaching gamma_d; certified alive stops at gamma_d - 1e-6)", - "parametric-identities-unproven-in-general (I1-I3 and Lemma C verified per member, 15 members; general (a,b) statement labelled SPECULATION)", - "a-greater-than-2b+3-uncovered (branch analysis breaks; W(6,3)-type members unmeasured)" - ], - "refuted_by": null - }, - { - "id": "004", - "file": "attempts/004-skeptic-review-of-003.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED_WITH_CORRECTIONS", - "mechanism": [ - "adversarial-verification", - "independent-reimplementation", - "symbolic-trig-closed-form", - "certified-interval-inequality" - ], - "one_line": "From-scratch re-verification confirms 003 end to end: glide reduction and mod-360 case tree re-derived by hand (1.9M-point adversarial sign search, zero dropped-branch hits), I1-I3 re-proven in an independent ring plus an off-torus grid-interpolation proof for W(2,1), Lemma C re-certified with an independent interval stack (4 members), all 8 exact alive certificates match digit-for-digit, four out-of-sample deaths re-measured to about 1e-12, mirror-half windows confirmed; two wording-level corrections, nothing load-bearing refuted.", - "leak_terms": [ - "dlsk_", - "off-torus grid", - "glide action p -> 2m - p", - "1938477" - ], - "range": "identities and glide facts re-proven for all 15 members (own ring; proof-grade off-torus grid for W(2,1); 20 exact spot points each); Lemma C re-certified for (2,1),(5,3),(10,9),(12,12); case tree searched at 1.94M adversarial points across all 15 members; all 8 exact alive certificates re-verified; deaths re-measured for (10,9),(12,12),(5,3),(4,2) plus calibration (2,1),(3,3); dead hunts at gamma_d + {0..0.5} deg for 4 members", - "verifies": "003", - "corrections": [ - "the W(5,3) exact alive certificate is at gamma_d - 1e-6, not the gamma_d - 1e-8 stated in 003 section on the general law (the Outcome/range/Death-brackets statements are correct; only that one sentence is wrong)", - "003 Outcome bullet corridor is empty at every gamma >= gamma_d overstates the theorem: it proves no positive width; a zero-width touching corridor at gamma_d is not excluded (alive is defined as positive width throughout, so nothing downstream is damaged)" - ], - "refuted_by": null, - "gaps": [] - }, - { - "id": "005", - "file": "attempts/005-complete-death-law-theorem.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED", - "mechanism": [ - "formal-parameter-specialization", - "symbolic-trig-closed-form", - "certified-interval-inequality", - "unfolding-word-census", - "out-of-sample-prediction" - ], - "one_line": "Death law completed on both sides: the half word composes to R_0 Rot_A(2a.alpha) Rot_B(-2b.beta), so I1-I4 and the glide facts become one exact 4-variable Laurent check valid for ALL (a,b); Lemmas C and D fall to an elementary c.cot(cs) monotonicity; the case tree is closed for all a >= b >= 1, a >= 2 (the a <= 2b+3 restriction removed); and certified alive segments into the death corner give death(W(a,b)) = gamma_d(a,b) EXACTLY (sup not attained) for 20 members including the newly measured W(6,3), W(6,1), W(8,2), W(8,1), W(7,1).", - "leak_terms": [ - "plaw_", - "R_0 Rot_A(2a alpha) Rot_B(-2b beta)", - "identity I4", - "margin N4", - "c cot(cs) monotone", - "Lemma L1", - "aw - bv = a(b+1)(theta - theta_d) - 360kb", - "rho = 2, t0 = 1/4", - "death corner (90/a, 90(a-1)/(a(b+1)))", - "585/4", - "255/2", - "285/2", - "1035/8", - "900/7", - "three-fold degenerate corner" - ], - "range": "I1-I4, D1, D2 and glide facts proven for ALL integers a, b >= 1 (formal ring; specialization cross-checked exactly on 23 members, floats to a = 40); Lemma C proven for all a >= 2, 1 <= b <= a and Lemma D for all b >= 1; necessity theorem for all integers a >= b >= 1, a >= 2 (no upper restriction on a); exact death = gamma_d certified for 20 members via alive segments t in (0, 1/4], gamma sweeping [gamma_d - 1/4, gamma_d); new float deaths for (6,3),(6,1),(8,2),(8,1),(7,1) at gamma_d to ~5e-12; adversarial case-tree scans ~1.56M points over 13 members", - "gaps": [ - "parametric-sufficiency-open", - "a-equals-1-column-unmeasured", - "birth-side-unproven" - ] - }, - { - "id": "006", - "file": "attempts/006-design-family-past-135.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED", - "mechanism": [ - "symbolic-trig-closed-form", - "unfolding-word-census", - "exact-rational-arithmetic", - "out-of-sample-prediction", - "structure-class-design-search" - ], - "one_line": "The 135 deg stall dissolves: 001 pinch gap was a sampler artifact (W(4,3) certified alive at gamma in (135.0000001, 135.048)), births obey a closed form gamma_birth = 180 - 90(a+b+1)/(a(b+1)) making staircase windows touch, the never-measured W(5,2) and a genuinely new four-block word are certified alive at gamma EXACTLY 135 (rational apexes on the arc, exact orbits), and the W family covers all 157 sampled obtuse arcs - while the design screens (3000+ canonical doubled words to length 50) find nothing of length <= 26 alive past 135 and show clean angle laws are confined to W's two-fan factorization.", - "leak_terms": [ - "design_", - "gamma_birth = 180 - 90(a+b+1)/(a(b+1))", - "W(5,2)", - "0(12)^3(02)^3(12)^2(02)^3", - "exactly 135", - "windows touch", - "(2x-1)^2 + (2y+1)^2 = 2", - "edge-accumulating sampler", - "136.8265" - ], - "range": "five exact certificates (rational apexes, Fraction corridors, independently simulated orbits): W(4,3) at certified gamma in (135.001,135.048) and (135.0000001,135.001); W(5,4) at (144.0000001,144.001); W(5,2) len 30 and non-W four-block N1 len 46 at gamma exactly 135 (exact on-arc identity); 003 obligations machine-certified for two new members (5,2),(6,2). Float screens sample-bounded: 245 canonical doubled words len <= 26 dead at 13 arcs >= 135; 811 len-30 words (only W(4,3),W(5,2) alive on 135-144); 2008 two-alternating words len 34-50 (only W members + two four-block words); three-block family a,b,c <= 3 (25/27 dead everywhere obtuse); birth law float-checked at 11 members; coverage checked at 157 arcs 90.5-165", - "gaps": [ - "birth-law-unproven (gamma_birth closed form is float-fitted + factor-structure motivated; no necessity/sufficiency theorem)", - "coverage-conjecture-open (union of windows covering (90,180) reduces to an elementary Diophantine statement, unproven)", - "length-26-stall-sample-bounded (doubled words only, 13 arcs; non-doubled even words 24-26 not rescreened)", - "four-block-death-angles-uncharacterized (binding differences provably non-elementary here only by failed factor search)" - ] - }, - { - "id": "007", - "file": "attempts/007-skeptic-review-of-006.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED_WITH_CORRECTIONS", - "mechanism": [ - "adversarial-verification", - "independent-reimplementation", - "exact-rational-arithmetic", - "out-of-sample-prediction" - ], - "one_line": "From-scratch re-verification confirms 006 everywhere it is exact: all five certificates re-established by a third corridor implementation (digit-for-digit widths) plus an own exact simulator, the exactly-135 on-arc identity re-derived by hand, 001's pinch-gap birth value regenerated from a replica of its own sampler (mechanism: the alive window is a corner sliver inside the grid's terminal 6e-4 clearance), the birth law survives three genuinely out-of-sample members to ~3e-11 (= sampler floor, including a > 2b+3), and the (5,2)/(6,2) theorem extension re-proves in an independent ring; four corrections, none load-bearing.", - "leak_terms": [ - "dsk_", - "566 not 811", - "canonical pairing (a,b) ~ (b+1,a-1)", - "corner sliver 6e-4", - "floor-tracking", - "W(8,2) birth 138.75" - ], - "range": "all 5 exact certificates re-verified (own matrix-form exact corridor, widths match digit-for-digit; own exact orbit simulation; own certified gamma brackets via 004's interval stack; on-arc gamma=135 identity hand-derived and Fraction-checked; Niven applicability checked); W(4,3) birth floor-tracked at 4 sampler depths; 001's census sampler replicated (reproduces 135.0486) and window x-extent measured at 7 gammas; birth law tested out-of-sample on W(7,3), W(6,3), W(8,2) (all +3.0e-11 = floor, dead-sampled below); coverage re-run at 25 own arcs (25/25 alive); universes recounted (245 and 2008 confirmed, 566-vs-811 correction); s1 hits reclassified (14 W + N1/N2); N1/N2 non-W checked incl. 0-1 swap; 60-word random dead re-test + 19 near-miss attacks (zero kills); identities + case tree (550k points) + Lemma C re-certified for (5,2),(6,2). NOT re-swept: full length-30 and s1 negative universes, three-block table, design_factor.py factorizations", - "verifies": "006", - "corrections": [ - "universe counts: the length-30 (|u|=15) universe is 566 canonical words, not 811 (811 is |u| <= 15 cumulative); 'all doubled odd words to length 30: 1056' double-counts (true count 811); the index one_line's '3000+' is 2846 counted correctly - screen coverage itself is unaffected", - "'length 30 is alive throughout the measured gap' overstates: certificates prove alive points AT gamma = 135 exactly and at one gamma in each of (135.0000001, 135.001) and (135.001, 135.048); the continuum statement is float evidence plus the SPECULATION window law", - "refutation semantics vs 001: 001's gap claim was knowledge-scoped ('nothing <= 30 KNOWN alive') and stays true as written; what 006 overturns is 001's birth measurements (error 0.0486 deg, 10x the stated ~0.005 sampling noise, whose 'deaths only under-estimated' hedge does not transfer to births) - correction/closure, not refutation", - "the window pairing (a,b) <-> (b+1, a-1) is a canonical-word identity (0<->1 swap + rotation/reversal; verified exactly on six pairs), so it is vacuous as evidence for the birth law, 006's 11-member fit spans only 9 canonical words, and 003's 'two distinct words dying on the same arc' (W(3,3)/W(4,2)) is a relabeling of one canonical word" - ], - "refuted_by": null, - "gaps": [] - }, - { - "id": "008", - "file": "attempts/008-skeptic-review-of-005.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED_WITH_CORRECTIONS", - "mechanism": [ - "adversarial-verification", - "independent-reimplementation", - "formal-parameter-specialization", - "symbolic-trig-closed-form", - "certified-interval-inequality" - ], - "one_line": "Adversarial re-verification confirms 005 end to end: the closed-form bridge u = R_0 Rot_A(2a.alpha) Rot_B(-2b.beta) recomposed exactly at off-torus rational points (40 members to a = 40), I1-I4 and glide facts re-proven in an independent 4-variable ring with u re-derived by symbolic map composition, Lemmas L1/C/D re-derived by hand and stressed to a = 10^6, the extended case tree re-derived branch by branch (threshold identity exact; 447k-point scan on 10 a > 2b+3 members, zero hits), the sufficiency segments independently certified alive by full-corridor interval unfolding down to gamma_d - 9.3e-10, and all five new deaths re-measured on the mirror half; one numeric correction, nothing load-bearing refuted.", - "leak_terms": [ - "psk_", - "off-torus pair algebra", - "gamma_d - 9.3e-10", - "447k", - "1e-4 not 1e-6" - ], - "range": "bridge recomposed exactly for 40 members incl. (40,40) at 2 rational points each; identities re-proven formally (my ring, 14/14); Lemma numerics 14,183 checks to a = 10^6; case tree scanned at ~447k adversarial points over 10 members with a > 2b+3 (k up to 2 reachable); segments certified alive at 42 points (7 members incl. the 5 new, t down to 2^-30); plaw_suffice trig layer audited (596 enclosure checks); 5 deaths re-measured with 004's sampler; plaw_suffice all NOT re-run end to end (selftest + one member reproduced, code audited)", - "verifies": "005", - "corrections": [ - "005 states the five new exact-alive certificates certify gamma within 1e-6 of gamma_d; the data files show gamma_d - 1e-4 for W(6,3), W(8,2), W(8,1), W(7,1) (only W(6,1) reaches 1e-6) - the --dg 1e-6 request is a float target, the certified bound is the largest 10^-k that certifies after apex snapping", - "caveat, not an error: the five new members' necessity rests on the parametric theorem alone (no per-member deathlaw_prove obligations were run for them, unlike 003's fifteen); sound by design since the identities are now proven for all (a,b)" - ], - "refuted_by": null, - "gaps": [] - }, - { - "id": "009", - "file": "attempts/009-lean-pilot-cot-lemma.md", - "date": "2026-08-04", - "mode": "informed", - "status": "FORMALIZED", - "mechanism": [ - "machine-checked-formalization", - "symbolic-trig-closed-form" - ], - "one_line": "First Lean 4 certificate in the lab: Lemma L1 of 005 (c.cot(cs) strictly decreasing on (0,1] for fixed s in (0, pi/2]) proved in mathlib with zero sorries and only the three standard axioms, statement matching 005 hypothesis-by-hypothesis with no narrowing; independent statement review and independent rebuild still pending.", - "leak_terms": [ - "cot_lemma_L1", - "strictAntiOn_id_mul_cot", - "BilliardsFormal", - "no HasDerivAt lemma for Real.cot", - "first Lean certificate" - ], - "verifies": "005", - "range": "exactly the formal statement StrictAntiOn (fun c => c * Real.cot (c * s)) (Ioc 0 1) for all real s with 0 < s <= pi/2 (Lean 4.32.2, mathlib v4.32.2); Lemmas C and D themselves, and the I1-I4 Laurent block, are NOT formalized", - "gaps": [ - "laurent-block-I1-I4-not-attempted (queue item 16 second half)" - ] - }, - { - "id": "010", - "file": "attempts/010-statement-review-of-009.md", - "date": "2026-08-04", - "mode": "informed", - "status": "VERIFIED", - "mechanism": [ - "adversarial-verification", - "machine-checked-formalization" - ], - "one_line": "Independent completion of the FORMALIZE.md pass on 009: non-formalizer statement review finds cot_lemma_L1 matches 005's Lemma L1 hypothesis-for-hypothesis with no narrowing (StrictAntiOn/Ioc/Real.cot read from vendored mathlib source, no junk-value reliance, cheat scan clean), own fresh lake build after deleting the project build output is green with axioms exactly [propext, Classical.choice, Quot.sound], and all three of 005's L1 use sites instantiate inside the formalized hypotheses; zero corrections, two cosmetic nits.", - "leak_terms": [ - "SkepticAxiomCheck", - "statement review signed off", - "AxiomCheck.lean not committed" - ], - "verifies": "009", - "range": "statement fidelity vs 005 section (ii)-B checked hypothesis by hypothesis; definitions Real.cot, StrictAntiOn, Set.Ioc read from mathlib rev 905b958 source; cheat scan over both project .lean files; rebuild = rm -rf .lake/build + lake build (8657 jobs, green) in a fresh container session with the pinned toolchain, mathlib oleans from the existing pinned cache (not re-downloaded); own axiom audit of both theorems; usage check of all three L1 call sites in 005. NOT audited: mathlib internals beyond the definitions used, Lean kernel, Lemmas C/D themselves", - "gaps": [] - }, - { - "id": "011", - "file": "attempts/011-lean-laurent-block.md", - "date": "2026-08-04", - "mode": "informed", - "status": "FORMALIZED", - "mechanism": [ - "machine-checked-formalization", - "formal-parameter-specialization" - ], - "one_line": "Second Lean 4 certificate: the full Laurent-identity block of 005 - closed-form half-word composition R0.RotA(2a alpha).RotB(-2b beta), I1-I4, D1, D2 and the glide facts - kernel-checked with (a,b) as formal exponent variables (23 theorems, zero sorries, standard three axioms only), tiers A+B+C green plus the letter-word iterate version for all natural (a,b); the block turns out homogeneous in i (no i^2 = -1 needed); geometry bridge out of scope per the lane, statement review pending.", - "leak_terms": [ - "LaurentBlock", - "half_word_letters", - "half_word_closed_form", - "homogeneous in i", - "pair encoding conj substitution", - "no AddMonoidAlgebra needed" - ], - "verifies": "005", - "range": "exactly the formal statements in formal/BilliardsFormal/LaurentBlock.lean: the identity block over an arbitrary field K with nonzero generators x, y, P, Q, i and 2 != 0 (P, Q carrying the (a,b) exponents formally; Tier C specializes P = x^a, Q = y^b for all integers a, b; half_word_letters covers the genuine letter word for all natural a, b); Lean 4.32.2, mathlib v4.32.2, axioms exactly [propext, Classical.choice, Quot.sound]", - "gaps": [ - "pending-independent-statement-review (encoding table vs 005 (ii)-A; FORMALIZE.md part 1)", - "pending-independent-rebuild (FORMALIZE.md part 3)", - "geometry-bridge-not-formalized (ring model = unfolding geometry carried by 005/008, out of scope by design)" - ] - } - ] -} \ No newline at end of file + "problem": "billiards-triangles", + "route_status": "EVIDENCE", + "route_summary": "Word-census route mapped blind to length 26: 376 exact certificates (family orbits to length 66, gamma <= 160 deg), coverage staircase stalling at 135 deg, and a measured hyperbolic (not super-exponential) growth of needed word length vs angle - but certified box width collapses exponentially in length, so length buys angle reach, not area. Negatives are sample-bounded; skeptic-confirmed with one numeric correction.", + "attempts": [ + { + "id": "001", + "file": "attempts/001-word-census-coverage-map.md", + "date": "2026-07-30", + "mode": "blind", + "status": "EVIDENCE", + "mechanism": [ + "unfolding-word-census", + "certified-box-coverage", + "exhaustive-enumeration" + ], + "one_line": "Canonical word census to length 26 with 376 exact certificates: coverage staircase stalls at 135 deg, blind rediscovery of the ~112.5 deg frontier, and a hyperbolic certificate-length growth law along the W(a,b) family (checked to length 66).", + "leak_terms": [ + "112.4989", + "W(a,b)", + "coverage staircase", + "length buys angle reach", + "17,527" + ], + "range": "canonical translation words of even length <= 26 (17,527 words); 1/64-cell midpoint screen of the obtuse half-disk; exact dyadic-box anchors; family rows to W(8,8), gamma <= 160 deg", + "gaps": [ + "sample-bounded-negatives", + "death-angle-laws-unproven" + ] + }, + { + "id": "002", + "file": "attempts/002-skeptic-review-of-001.md", + "date": "2026-07-30", + "mode": "blind", + "status": "VERIFIED_WITH_CORRECTIONS", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "unfolding-word-census" + ], + "one_line": "From-scratch re-implementation of every layer confirms 001 (36/36 sampled certificates at 180 exact apexes, 17,527-word universe reproduced word-for-word, death law survives two out-of-sample prediction rows); certified-area figure corrected to ~4.3e-6.", + "leak_terms": [ + "147.5974", + "157.4976", + "4.3e-6" + ], + "verifies": "001", + "corrections": [ + "certified-area total over all 376 anchors is ~4.3e-6, not ~2.2e-6 (which is the map-anchor subtotal)", + "the 64.3% uncovered figure is of the resolved interior, not the whole obtuse region", + "anchor simulation re-derivation is at the anchor centre; the full 9-sample box attack ran on every 5th/7th record", + "kill-condition verdict is measured only to gamma ~ 160 deg; beyond that the hyperbolic form is the SPECULATION law's extrapolation" + ] + }, + { + "id": "003", + "file": "attempts/003-death-angle-laws.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED", + "mechanism": [ + "symbolic-trig-closed-form", + "certified-interval-inequality", + "out-of-sample-prediction", + "unfolding-word-census" + ], + "one_line": "The W(a,b) death-angle laws of 001 survive both out-of-sample tests (W(10,9) -> 162.90, W(12,12) -> 2160/13, to ~1e-12), and are then derived: exact division-free unfolding gives closed-form binding functions, a general law gamma_d = 180 - 90(a+b)/(a(b+1)) (confirmed out-of-sample on W(5,3) -> 144 and W(4,2) -> 135), and a machine-certified necessity theorem: the corridor is empty at every gamma >= gamma_d, for 15 verified members.", + "leak_terms": [ + "gamma_d = 180 - 90(a+b)/(a(b+1))", + "90(a+b)/(a(b+1))", + "glide reflection reduction", + "W(5,3)", + "144.0000000000", + "deathlaw_", + "Lemma C", + "division-free unfolding" + ], + "range": "necessity theorem machine-certified for the 15 members (2,1),(2,2),(3,2),(3,3),(4,2),(4,3),(4,4),(5,3),(5,4),(5,5),(6,6),(7,7),(8,8),(10,9),(12,12); exact alive certificates at gamma_d - 1e-6 (1e-8 for (3,3),(4,2)) for 8 members; float deaths measured to ~1e-11 for 11 members; parametric statement for all a >= b >= 1, a >= 2, a <= 2b+3 rests on two per-member-verified inputs (identities I1-I3 and Lemma C) labelled SPECULATION in general", + "gaps": [ + "sufficiency-limit-unclosed (death = gamma_d exactly needs alive points approaching gamma_d; certified alive stops at gamma_d - 1e-6)", + "parametric-identities-unproven-in-general (I1-I3 and Lemma C verified per member, 15 members; general (a,b) statement labelled SPECULATION)", + "a-greater-than-2b+3-uncovered (branch analysis breaks; W(6,3)-type members unmeasured)" + ], + "refuted_by": null + }, + { + "id": "004", + "file": "attempts/004-skeptic-review-of-003.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED_WITH_CORRECTIONS", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "symbolic-trig-closed-form", + "certified-interval-inequality" + ], + "one_line": "From-scratch re-verification confirms 003 end to end: glide reduction and mod-360 case tree re-derived by hand (1.9M-point adversarial sign search, zero dropped-branch hits), I1-I3 re-proven in an independent ring plus an off-torus grid-interpolation proof for W(2,1), Lemma C re-certified with an independent interval stack (4 members), all 8 exact alive certificates match digit-for-digit, four out-of-sample deaths re-measured to about 1e-12, mirror-half windows confirmed; two wording-level corrections, nothing load-bearing refuted.", + "leak_terms": [ + "dlsk_", + "off-torus grid", + "glide action p -> 2m - p", + "1938477" + ], + "range": "identities and glide facts re-proven for all 15 members (own ring; proof-grade off-torus grid for W(2,1); 20 exact spot points each); Lemma C re-certified for (2,1),(5,3),(10,9),(12,12); case tree searched at 1.94M adversarial points across all 15 members; all 8 exact alive certificates re-verified; deaths re-measured for (10,9),(12,12),(5,3),(4,2) plus calibration (2,1),(3,3); dead hunts at gamma_d + {0..0.5} deg for 4 members", + "verifies": "003", + "corrections": [ + "the W(5,3) exact alive certificate is at gamma_d - 1e-6, not the gamma_d - 1e-8 stated in 003 section on the general law (the Outcome/range/Death-brackets statements are correct; only that one sentence is wrong)", + "003 Outcome bullet corridor is empty at every gamma >= gamma_d overstates the theorem: it proves no positive width; a zero-width touching corridor at gamma_d is not excluded (alive is defined as positive width throughout, so nothing downstream is damaged)" + ], + "refuted_by": null, + "gaps": [] + }, + { + "id": "005", + "file": "attempts/005-complete-death-law-theorem.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED", + "mechanism": [ + "formal-parameter-specialization", + "symbolic-trig-closed-form", + "certified-interval-inequality", + "unfolding-word-census", + "out-of-sample-prediction" + ], + "one_line": "Death law completed on both sides: the half word composes to R_0 Rot_A(2a.alpha) Rot_B(-2b.beta), so I1-I4 and the glide facts become one exact 4-variable Laurent check valid for ALL (a,b); Lemmas C and D fall to an elementary c.cot(cs) monotonicity; the case tree is closed for all a >= b >= 1, a >= 2 (the a <= 2b+3 restriction removed); and certified alive segments into the death corner give death(W(a,b)) = gamma_d(a,b) EXACTLY (sup not attained) for 20 members including the newly measured W(6,3), W(6,1), W(8,2), W(8,1), W(7,1).", + "leak_terms": [ + "plaw_", + "R_0 Rot_A(2a alpha) Rot_B(-2b beta)", + "identity I4", + "margin N4", + "c cot(cs) monotone", + "Lemma L1", + "aw - bv = a(b+1)(theta - theta_d) - 360kb", + "rho = 2, t0 = 1/4", + "death corner (90/a, 90(a-1)/(a(b+1)))", + "585/4", + "255/2", + "285/2", + "1035/8", + "900/7", + "three-fold degenerate corner" + ], + "range": "I1-I4, D1, D2 and glide facts proven for ALL integers a, b >= 1 (formal ring; specialization cross-checked exactly on 23 members, floats to a = 40); Lemma C proven for all a >= 2, 1 <= b <= a and Lemma D for all b >= 1; necessity theorem for all integers a >= b >= 1, a >= 2 (no upper restriction on a); exact death = gamma_d certified for 20 members via alive segments t in (0, 1/4], gamma sweeping [gamma_d - 1/4, gamma_d); new float deaths for (6,3),(6,1),(8,2),(8,1),(7,1) at gamma_d to ~5e-12; adversarial case-tree scans ~1.56M points over 13 members", + "gaps": [ + "parametric-sufficiency-open", + "a-equals-1-column-unmeasured", + "birth-side-unproven" + ] + }, + { + "id": "006", + "file": "attempts/006-design-family-past-135.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED", + "mechanism": [ + "symbolic-trig-closed-form", + "unfolding-word-census", + "exact-rational-arithmetic", + "out-of-sample-prediction", + "structure-class-design-search" + ], + "one_line": "The 135 deg stall dissolves: 001 pinch gap was a sampler artifact (W(4,3) certified alive at gamma in (135.0000001, 135.048)), births obey a closed form gamma_birth = 180 - 90(a+b+1)/(a(b+1)) making staircase windows touch, the never-measured W(5,2) and a genuinely new four-block word are certified alive at gamma EXACTLY 135 (rational apexes on the arc, exact orbits), and the W family covers all 157 sampled obtuse arcs - while the design screens (3000+ canonical doubled words to length 50) find nothing of length <= 26 alive past 135 and show clean angle laws are confined to W's two-fan factorization.", + "leak_terms": [ + "design_", + "gamma_birth = 180 - 90(a+b+1)/(a(b+1))", + "W(5,2)", + "0(12)^3(02)^3(12)^2(02)^3", + "exactly 135", + "windows touch", + "(2x-1)^2 + (2y+1)^2 = 2", + "edge-accumulating sampler", + "136.8265" + ], + "range": "five exact certificates (rational apexes, Fraction corridors, independently simulated orbits): W(4,3) at certified gamma in (135.001,135.048) and (135.0000001,135.001); W(5,4) at (144.0000001,144.001); W(5,2) len 30 and non-W four-block N1 len 46 at gamma exactly 135 (exact on-arc identity); 003 obligations machine-certified for two new members (5,2),(6,2). Float screens sample-bounded: 245 canonical doubled words len <= 26 dead at 13 arcs >= 135; 811 len-30 words (only W(4,3),W(5,2) alive on 135-144); 2008 two-alternating words len 34-50 (only W members + two four-block words); three-block family a,b,c <= 3 (25/27 dead everywhere obtuse); birth law float-checked at 11 members; coverage checked at 157 arcs 90.5-165", + "gaps": [ + "birth-law-unproven (gamma_birth closed form is float-fitted + factor-structure motivated; no necessity/sufficiency theorem)", + "coverage-conjecture-open (union of windows covering (90,180) reduces to an elementary Diophantine statement, unproven)", + "length-26-stall-sample-bounded (doubled words only, 13 arcs; non-doubled even words 24-26 not rescreened)", + "four-block-death-angles-uncharacterized (binding differences provably non-elementary here only by failed factor search)" + ] + }, + { + "id": "007", + "file": "attempts/007-skeptic-review-of-006.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED_WITH_CORRECTIONS", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "exact-rational-arithmetic", + "out-of-sample-prediction" + ], + "one_line": "From-scratch re-verification confirms 006 everywhere it is exact: all five certificates re-established by a third corridor implementation (digit-for-digit widths) plus an own exact simulator, the exactly-135 on-arc identity re-derived by hand, 001's pinch-gap birth value regenerated from a replica of its own sampler (mechanism: the alive window is a corner sliver inside the grid's terminal 6e-4 clearance), the birth law survives three genuinely out-of-sample members to ~3e-11 (= sampler floor, including a > 2b+3), and the (5,2)/(6,2) theorem extension re-proves in an independent ring; four corrections, none load-bearing.", + "leak_terms": [ + "dsk_", + "566 not 811", + "canonical pairing (a,b) ~ (b+1,a-1)", + "corner sliver 6e-4", + "floor-tracking", + "W(8,2) birth 138.75" + ], + "range": "all 5 exact certificates re-verified (own matrix-form exact corridor, widths match digit-for-digit; own exact orbit simulation; own certified gamma brackets via 004's interval stack; on-arc gamma=135 identity hand-derived and Fraction-checked; Niven applicability checked); W(4,3) birth floor-tracked at 4 sampler depths; 001's census sampler replicated (reproduces 135.0486) and window x-extent measured at 7 gammas; birth law tested out-of-sample on W(7,3), W(6,3), W(8,2) (all +3.0e-11 = floor, dead-sampled below); coverage re-run at 25 own arcs (25/25 alive); universes recounted (245 and 2008 confirmed, 566-vs-811 correction); s1 hits reclassified (14 W + N1/N2); N1/N2 non-W checked incl. 0-1 swap; 60-word random dead re-test + 19 near-miss attacks (zero kills); identities + case tree (550k points) + Lemma C re-certified for (5,2),(6,2). NOT re-swept: full length-30 and s1 negative universes, three-block table, design_factor.py factorizations", + "verifies": "006", + "corrections": [ + "universe counts: the length-30 (|u|=15) universe is 566 canonical words, not 811 (811 is |u| <= 15 cumulative); 'all doubled odd words to length 30: 1056' double-counts (true count 811); the index one_line's '3000+' is 2846 counted correctly - screen coverage itself is unaffected", + "'length 30 is alive throughout the measured gap' overstates: certificates prove alive points AT gamma = 135 exactly and at one gamma in each of (135.0000001, 135.001) and (135.001, 135.048); the continuum statement is float evidence plus the SPECULATION window law", + "refutation semantics vs 001: 001's gap claim was knowledge-scoped ('nothing <= 30 KNOWN alive') and stays true as written; what 006 overturns is 001's birth measurements (error 0.0486 deg, 10x the stated ~0.005 sampling noise, whose 'deaths only under-estimated' hedge does not transfer to births) - correction/closure, not refutation", + "the window pairing (a,b) <-> (b+1, a-1) is a canonical-word identity (0<->1 swap + rotation/reversal; verified exactly on six pairs), so it is vacuous as evidence for the birth law, 006's 11-member fit spans only 9 canonical words, and 003's 'two distinct words dying on the same arc' (W(3,3)/W(4,2)) is a relabeling of one canonical word" + ], + "refuted_by": null, + "gaps": [] + }, + { + "id": "008", + "file": "attempts/008-skeptic-review-of-005.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED_WITH_CORRECTIONS", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "formal-parameter-specialization", + "symbolic-trig-closed-form", + "certified-interval-inequality" + ], + "one_line": "Adversarial re-verification confirms 005 end to end: the closed-form bridge u = R_0 Rot_A(2a.alpha) Rot_B(-2b.beta) recomposed exactly at off-torus rational points (40 members to a = 40), I1-I4 and glide facts re-proven in an independent 4-variable ring with u re-derived by symbolic map composition, Lemmas L1/C/D re-derived by hand and stressed to a = 10^6, the extended case tree re-derived branch by branch (threshold identity exact; 447k-point scan on 10 a > 2b+3 members, zero hits), the sufficiency segments independently certified alive by full-corridor interval unfolding down to gamma_d - 9.3e-10, and all five new deaths re-measured on the mirror half; one numeric correction, nothing load-bearing refuted.", + "leak_terms": [ + "psk_", + "off-torus pair algebra", + "gamma_d - 9.3e-10", + "447k", + "1e-4 not 1e-6" + ], + "range": "bridge recomposed exactly for 40 members incl. (40,40) at 2 rational points each; identities re-proven formally (my ring, 14/14); Lemma numerics 14,183 checks to a = 10^6; case tree scanned at ~447k adversarial points over 10 members with a > 2b+3 (k up to 2 reachable); segments certified alive at 42 points (7 members incl. the 5 new, t down to 2^-30); plaw_suffice trig layer audited (596 enclosure checks); 5 deaths re-measured with 004's sampler; plaw_suffice all NOT re-run end to end (selftest + one member reproduced, code audited)", + "verifies": "005", + "corrections": [ + "005 states the five new exact-alive certificates certify gamma within 1e-6 of gamma_d; the data files show gamma_d - 1e-4 for W(6,3), W(8,2), W(8,1), W(7,1) (only W(6,1) reaches 1e-6) - the --dg 1e-6 request is a float target, the certified bound is the largest 10^-k that certifies after apex snapping", + "caveat, not an error: the five new members' necessity rests on the parametric theorem alone (no per-member deathlaw_prove obligations were run for them, unlike 003's fifteen); sound by design since the identities are now proven for all (a,b)" + ], + "refuted_by": null, + "gaps": [] + }, + { + "id": "009", + "file": "attempts/009-lean-pilot-cot-lemma.md", + "date": "2026-08-04", + "mode": "informed", + "status": "FORMALIZED", + "mechanism": [ + "machine-checked-formalization", + "symbolic-trig-closed-form" + ], + "one_line": "First Lean 4 certificate in the lab: Lemma L1 of 005 (c.cot(cs) strictly decreasing on (0,1] for fixed s in (0, pi/2]) proved in mathlib with zero sorries and only the three standard axioms, statement matching 005 hypothesis-by-hypothesis with no narrowing; independent statement review and independent rebuild still pending.", + "leak_terms": [ + "cot_lemma_L1", + "strictAntiOn_id_mul_cot", + "BilliardsFormal", + "no HasDerivAt lemma for Real.cot", + "first Lean certificate" + ], + "verifies": "005", + "range": "exactly the formal statement StrictAntiOn (fun c => c * Real.cot (c * s)) (Ioc 0 1) for all real s with 0 < s <= pi/2 (Lean 4.32.2, mathlib v4.32.2); Lemmas C and D themselves, and the I1-I4 Laurent block, are NOT formalized", + "gaps": [ + "laurent-block-I1-I4-not-attempted (queue item 16 second half)" + ] + }, + { + "id": "010", + "file": "attempts/010-statement-review-of-009.md", + "date": "2026-08-04", + "mode": "informed", + "status": "VERIFIED", + "mechanism": [ + "adversarial-verification", + "machine-checked-formalization" + ], + "one_line": "Independent completion of the FORMALIZE.md pass on 009: non-formalizer statement review finds cot_lemma_L1 matches 005's Lemma L1 hypothesis-for-hypothesis with no narrowing (StrictAntiOn/Ioc/Real.cot read from vendored mathlib source, no junk-value reliance, cheat scan clean), own fresh lake build after deleting the project build output is green with axioms exactly [propext, Classical.choice, Quot.sound], and all three of 005's L1 use sites instantiate inside the formalized hypotheses; zero corrections, two cosmetic nits.", + "leak_terms": [ + "SkepticAxiomCheck", + "statement review signed off", + "AxiomCheck.lean not committed" + ], + "verifies": "009", + "range": "statement fidelity vs 005 section (ii)-B checked hypothesis by hypothesis; definitions Real.cot, StrictAntiOn, Set.Ioc read from mathlib rev 905b958 source; cheat scan over both project .lean files; rebuild = rm -rf .lake/build + lake build (8657 jobs, green) in a fresh container session with the pinned toolchain, mathlib oleans from the existing pinned cache (not re-downloaded); own axiom audit of both theorems; usage check of all three L1 call sites in 005. NOT audited: mathlib internals beyond the definitions used, Lean kernel, Lemmas C/D themselves", + "gaps": [] + }, + { + "id": "011", + "file": "attempts/011-lean-laurent-block.md", + "date": "2026-08-04", + "mode": "informed", + "status": "FORMALIZED", + "mechanism": [ + "machine-checked-formalization", + "formal-parameter-specialization" + ], + "one_line": "Second Lean 4 certificate: the full Laurent-identity block of 005 - closed-form half-word composition R0.RotA(2a alpha).RotB(-2b beta), I1-I4, D1, D2 and the glide facts - kernel-checked with (a,b) as formal exponent variables (23 theorems, zero sorries, standard three axioms only), tiers A+B+C green plus the letter-word iterate version for all natural (a,b); the block turns out homogeneous in i (no i^2 = -1 needed); geometry bridge out of scope per the lane, statement review pending.", + "leak_terms": [ + "LaurentBlock", + "half_word_letters", + "half_word_closed_form", + "homogeneous in i", + "pair encoding conj substitution", + "no AddMonoidAlgebra needed" + ], + "verifies": "005", + "range": "exactly the formal statements in formal/BilliardsFormal/LaurentBlock.lean: the identity block over an arbitrary field K with nonzero generators x, y, P, Q, i and 2 != 0 (P, Q carrying the (a,b) exponents formally; Tier C specializes P = x^a, Q = y^b for all integers a, b; half_word_letters covers the genuine letter word for all natural a, b); Lean 4.32.2, mathlib v4.32.2, axioms exactly [propext, Classical.choice, Quot.sound]", + "gaps": [ + "geometry-bridge-not-formalized (ring model = unfolding geometry carried by 005/008, out of scope by design)" + ] + }, + { + "id": "012", + "file": "attempts/012-statement-review-of-011.md", + "date": "2026-08-04", + "mode": "informed", + "status": "VERIFIED", + "mechanism": [ + "adversarial-verification", + "machine-checked-formalization", + "independent-reimplementation", + "formal-parameter-specialization" + ], + "one_line": "Independent completion of the FORMALIZE.md pass on 011: non-formalizer statement review confirms all 23 Lean theorems say what 005's Laurent block says (definitions, angle dictionary, signs, conj-as-substitution = 008's star involution, pair encoding assumption-free, i-drop safe, no nat-vs-int narrowing, v2's Q genuinely absent in 005 too); every theorem re-proven in an own 7-variable symbolic ring with i formal (31/31 exact zero polynomials, independently confirming the i-homogeneity), D1 additionally re-derived by hand, statements matched to 005's prose at 200 random points to ~4e-15; own fresh lake build green, own axiom audit of all 23 theorems exactly the standard three; zero corrections, three cosmetic nits.", + "leak_terms": [ + "lbsk_", + "7-variable ring i formal", + "31/31 zero polynomials", + "rationality invariant of the conj substitution", + "Laurent block pass complete" + ], + "verifies": "011", + "range": "statement fidelity of all 23 theorems in formal/BilliardsFormal/LaurentBlock.lean checked against 005 (ii)-A Steps 2-4 and plaw_general.closed_forms definition by definition; every theorem re-proven symbolically in an independent stdlib ring with i as a formal Laurent variable (half_word_letters re-composed by symbolic iteration for a,b <= 3); Lean-to-005 mapping float-checked at 200 random (a,b,alpha,beta), a <= 12, 30 quantities; cheat scan over all four project .lean files; rebuild = rm -rf .lake/build + lake build (8658 jobs, green, LaurentBlock re-elaborated 41 s) in a fresh session with the pinned toolchain and mathlib cache; own axiom audit of all 23 theorems from an own scratch file. NOT audited: the geometry bridge (permanent certificate scope, carried by 005/008), Lemmas C/D and the analytic layer, mathlib internals, the Lean kernel", + "gaps": [] + } + ] +} From 0e3ccbec7329dcddf98e0a91787ada4ecb845b61 Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 10:38:23 +0000 Subject: [PATCH 09/14] union-closed 014: aggregated-control probe to n=160 - KILL claimed, skeptic pending Queue item 1's probe-first step: the i-aggregated odds-ratio control survives every seeded attack at n <= 32 (013 anchor reproduced exact), but a replicated-witness ladder MU(n,r) with orbit-optimized shared unit weights drives the aggregate negative past n ~ 90: certified in exact rational arithmetic at t=4 (lambda=2) at n=96/128/160 with marginals <= 0.309 < 0.38271, enclosure widths <= 1e-18, dichotomy checks exact, sign robust to the alternative bookkeeping and 3% perturbations. Claimed ledger consequence (pending skeptic): the aggregated-control proof branch closes negative; the gap falls to the margin-modulated control and a lambda <~ c/n restricted variant. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- .../014-aggregated-control-probe-large-n.md | 383 +++++++++ problems/union-closed/data/aggprobe2_kill.log | 19 + .../union-closed/data/aggprobe2_n96cert.json | 14 + .../union-closed/data/aggprobe2_n96cert.log | 15 + .../data/aggprobe2_p3best_cert.json | 13 + .../union-closed/data/aggprobe2_partP1.json | 172 +++++ .../union-closed/data/aggprobe2_partP2.json | 148 ++++ .../union-closed/data/aggprobe2_partP3.json | 135 ++++ .../union-closed/data/aggprobe2_partP4.json | 116 +++ .../union-closed/data/aggprobe2_partP5.json | 462 +++++++++++ .../union-closed/data/aggprobe2_partP6.json | 202 +++++ .../union-closed/data/aggprobe2_partP7.json | 563 ++++++++++++++ problems/union-closed/data/aggprobe2_run.log | 35 + .../union-closed/data/aggprobe2_sweep.log | 25 + .../union-closed/data/aggprobe_partA.json | 122 +++ .../union-closed/data/aggprobe_partB.json | 724 ++++++++++++++++++ .../union-closed/data/aggprobe_partC_n12.json | 155 ++++ .../union-closed/data/aggprobe_partC_n16.json | 160 ++++ .../union-closed/data/aggprobe_partC_n20.json | 168 ++++ .../union-closed/data/aggprobe_partC_n24.json | 167 ++++ .../union-closed/data/aggprobe_partC_n28.json | 165 ++++ .../union-closed/data/aggprobe_partC_n32.json | 164 ++++ .../union-closed/data/aggprobe_partC_n8.json | 126 +++ .../data/aggprobe_partC_summary.json | 67 ++ .../union-closed/data/aggprobe_partD.json | 49 ++ .../union-closed/data/aggprobe_partE.json | 69 ++ problems/union-closed/data/aggprobe_run.log | 40 + .../union-closed/explore/uc_or_agg_probe.py | 594 ++++++++++++++ .../union-closed/explore/uc_or_agg_probe2.py | 631 +++++++++++++++ problems/union-closed/prior-art.json | 38 +- 30 files changed, 5740 insertions(+), 1 deletion(-) create mode 100644 problems/union-closed/attempts/014-aggregated-control-probe-large-n.md create mode 100644 problems/union-closed/data/aggprobe2_kill.log create mode 100644 problems/union-closed/data/aggprobe2_n96cert.json create mode 100644 problems/union-closed/data/aggprobe2_n96cert.log create mode 100644 problems/union-closed/data/aggprobe2_p3best_cert.json create mode 100644 problems/union-closed/data/aggprobe2_partP1.json create mode 100644 problems/union-closed/data/aggprobe2_partP2.json create mode 100644 problems/union-closed/data/aggprobe2_partP3.json create mode 100644 problems/union-closed/data/aggprobe2_partP4.json create mode 100644 problems/union-closed/data/aggprobe2_partP5.json create mode 100644 problems/union-closed/data/aggprobe2_partP6.json create mode 100644 problems/union-closed/data/aggprobe2_partP7.json create mode 100644 problems/union-closed/data/aggprobe2_run.log create mode 100644 problems/union-closed/data/aggprobe2_sweep.log create mode 100644 problems/union-closed/data/aggprobe_partA.json create mode 100644 problems/union-closed/data/aggprobe_partB.json create mode 100644 problems/union-closed/data/aggprobe_partC_n12.json create mode 100644 problems/union-closed/data/aggprobe_partC_n16.json create mode 100644 problems/union-closed/data/aggprobe_partC_n20.json create mode 100644 problems/union-closed/data/aggprobe_partC_n24.json create mode 100644 problems/union-closed/data/aggprobe_partC_n28.json create mode 100644 problems/union-closed/data/aggprobe_partC_n32.json create mode 100644 problems/union-closed/data/aggprobe_partC_n8.json create mode 100644 problems/union-closed/data/aggprobe_partC_summary.json create mode 100644 problems/union-closed/data/aggprobe_partD.json create mode 100644 problems/union-closed/data/aggprobe_partE.json create mode 100644 problems/union-closed/data/aggprobe_run.log create mode 100644 problems/union-closed/explore/uc_or_agg_probe.py create mode 100644 problems/union-closed/explore/uc_or_agg_probe2.py diff --git a/problems/union-closed/attempts/014-aggregated-control-probe-large-n.md b/problems/union-closed/attempts/014-aggregated-control-probe-large-n.md new file mode 100644 index 0000000..8c0c049 --- /dev/null +++ b/problems/union-closed/attempts/014-aggregated-control-probe-large-n.md @@ -0,0 +1,383 @@ +# 014 — Large-n probe kills the i-aggregated odds-ratio control + +- **Problem:** union-closed, `problems/union-closed/PROBLEM.md` +- **Date:** 2026-08-04 +- **Mode:** informed +- **Type:** computational search (falsifiable probe of the restated Gap 1 at + n ≫ 7, run BEFORE proof effort per queue item 1); adversarial families + + exact rational certification. Outcome: the aggregated control is **REFUTED + in the record-relevant regime** by an explicit unit-replicated family at + n ≈ 96–128, certified float-free. +- **Tools:** `explore/uc_or_agg_probe.py` (round 1: anchors, ladder trend to + n = 32, seeded free-support climbs, permutation attack, exact certs; + ~3 min) and `explore/uc_or_agg_probe2.py` (round 2: ladder extension to + n = 128, unit-symmetric θ-optimization, deep annealed climbs, adversarial + dilution sweep `--sweep`, kill drill `--kill`; ~30 min + drill). + Both standard-library only, deterministic (fixed seeds, fixed step counts). + Float census is 007's engine (`uc_or_avg.py`) imported verbatim; exact + rational census + certified log₂ enclosures are 013's engine + (`uc_or_avg_skeptic.py`) imported verbatim; a second float engine + (013's own `f_census`) is used as an independent cross-check on the + violating instance. Commands reproducing every number: + + python problems/union-closed/explore/uc_or_agg_probe.py 2>&1 | tee problems/union-closed/data/aggprobe_run.log + python problems/union-closed/explore/uc_or_agg_probe2.py 2>&1 | tee problems/union-closed/data/aggprobe2_run.log + python problems/union-closed/explore/uc_or_agg_probe2.py --sweep 2>&1 | tee problems/union-closed/data/aggprobe2_sweep.log + python problems/union-closed/explore/uc_or_agg_probe2.py --kill 2>&1 | tee problems/union-closed/data/aggprobe2_kill.log + python problems/union-closed/explore/uc_or_agg_probe2.py --cert96 2>&1 | tee problems/union-closed/data/aggprobe2_n96cert.log + python problems/union-closed/explore/uc_or_agg_probe2.py --certp3 # re-cert of P3-best, writes aggprobe2_p3best_cert.json + + Checkpoints: `data/aggprobe_part[ABDE]*.json`, `data/aggprobe_partC_n*.json`, + `data/aggprobe2_partP[1-7]*.json`. +- **Sources:** 007/013 (definition of M_i, the aggregate, the 10-atom + witness, exact-arithmetic standard), 012 (large-n reversal warning, orbit + symmetrization as the way to buy n), 005/006 (families), STATUS.md queue + item 1. No external fetches this cycle. + +## Approach + +**The statement under test** (007 lead 1 = 013 claim 6, the restated Gap 1 +`aggregated-or-control`; stated here exactly as certified positive in 013 +part C). For μ on 2^[n] with H(μ) > 0 and λ > 0, with M_i(μ, λ) the +well-posed history-averaged control of 007 §1 (mass-weighted mean of +log₂ OR_i(a,b) over the nondegenerate history block N_i × N_i of the tilt +coupling π_λ ∝ u(A)u(B)2^{λ|A∩B|}) and w_i = the defined history mass at +coordinate i: + + A(μ, λ) = Σ_{i= 0 expected) +[B] MU ladder lam=2.0 full: agg vs n: 7:+1.3525 10:+1.2621 13:+1.1806 16:+1.1067 19:+1.0395 22:+0.9783 25:+0.9222 28:+0.8708 32:+0.8087 +[B] per-i min vs n: 7:-0.0401 10:-0.0401 13:-0.0401 16:-0.0401 19:-0.0401 22:-0.0401 25:-0.0401 28:-0.0401 32:-0.0401 +[B] max marginal vs n: 7:0.284 10:0.282 13:0.281 16:0.280 19:0.279 22:0.278 25:0.277 28:0.276 32:0.274 +[B] MU ladder lam=2.0 fixed6: agg vs n: 13:+1.2068 16:+1.2068 19:+1.2068 22:+1.2068 25:+1.2068 28:+1.2068 32:+1.2068 +[B] per-i min vs n: 13:-0.0401 16:-0.0401 19:-0.0401 22:-0.0401 25:-0.0401 28:-0.0401 32:-0.0401 +[B] max marginal vs n: 13:0.282 16:0.282 19:0.282 22:0.282 25:0.282 28:0.282 32:0.282 +[B] MU ladder lam=3.5 full: agg vs n: 7:+2.5996 10:+2.3434 13:+2.1264 16:+1.9405 19:+1.7797 22:+1.6392 25:+1.5157 28:+1.4064 32:+1.2789 +[B] per-i min vs n: 7:-0.1220 10:-0.1220 13:-0.1220 16:-0.1220 19:-0.1220 22:-0.1220 25:-0.1220 28:-0.1220 32:-0.1220 +[B] max marginal vs n: 7:0.286 10:0.285 13:0.284 16:0.283 19:0.282 22:0.280 25:0.279 28:0.278 32:0.276 +[B] MU ladder lam=3.5 fixed6: agg vs n: 13:+2.1950 16:+2.1950 19:+2.1950 22:+2.1950 25:+2.1950 28:+2.1950 32:+2.1950 +[B] per-i min vs n: 13:-0.1220 16:-0.1220 19:-0.1220 22:-0.1220 25:-0.1220 28:-0.1220 32:-0.1220 +[B] max marginal vs n: 13:0.284 16:0.284 19:0.284 22:0.284 25:0.284 28:0.284 32:0.284 +[B] MU ladder lam=5.0 full: agg vs n: 7:+3.7650 10:+3.1894 13:+2.7578 16:+2.4224 19:+2.1545 22:+1.9358 25:+1.7540 28:+1.6006 32:+1.4299 +[B] per-i min vs n: 7:-0.0780 10:-0.0780 13:-0.0780 16:-0.0780 19:-0.0780 22:-0.0780 25:-0.0780 28:-0.0780 32:-0.0780 +[B] max marginal vs n: 7:0.269 10:0.268 13:0.267 16:0.266 19:0.265 22:0.263 25:0.262 28:0.261 32:0.259 +[B] MU ladder lam=5.0 fixed6: agg vs n: 13:+2.8891 16:+2.8891 19:+2.8891 22:+2.8891 25:+2.8891 28:+2.8891 32:+2.8891 +[B] per-i min vs n: 13:-0.0780 16:-0.0780 19:-0.0780 22:-0.0780 25:-0.0780 28:-0.0780 32:-0.0780 +[B] max marginal vs n: 13:0.267 16:0.267 19:0.267 22:0.267 25:0.267 28:0.267 32:0.267 +[C] n=8: 7 climbs x 500 steps, in-regime aggregate violations: 0; three lowest endpoints: +0.00406 +0.01401 +0.16611 (deepest per-i min -0.0290) +[C] n=12: 8 climbs x 500 steps, in-regime aggregate violations: 0; three lowest endpoints: +0.15688 +0.35391 +0.76562 (deepest per-i min -0.1203) +[C] n=16: 8 climbs x 500 steps, in-regime aggregate violations: 0; three lowest endpoints: +0.26185 +0.42299 +0.55815 (deepest per-i min -0.1076) +[C] n=20: 8 climbs x 400 steps, in-regime aggregate violations: 0; three lowest endpoints: +0.29210 +0.31462 +0.58902 (deepest per-i min -0.1037) +[C] n=24: 8 climbs x 400 steps, in-regime aggregate violations: 0; three lowest endpoints: +0.33506 +0.43785 +0.72213 (deepest per-i min -0.1207) +[C] n=28: 8 climbs x 300 steps, in-regime aggregate violations: 0; three lowest endpoints: +0.33766 +0.44225 +1.12057 (deepest per-i min -0.1171) +[C] n=32: 8 climbs x 300 steps, in-regime aggregate violations: 0; three lowest endpoints: +0.16314 +0.28032 +0.42340 (deepest per-i min -0.0996) +[D] permutation attack witness_n7: base +1.8448, 41 orders, min +1.6939, in-regime violations 0 +[D] permutation attack MU(13): base +2.1264, 41 orders, min +2.2000, in-regime violations 0 +[D] permutation attack MU(22): base +1.7308, 41 orders, min +1.8986, in-regime violations 0 +[D] permutation attack MU(32): base +1.7308, 41 orders, min +1.8292, in-regime violations 0 +[D] permutation attack search_best_n8: base +0.1661, 41 orders, min +0.0183, in-regime violations 0 +[E] exact witness_n7_t181/16: aggregate in [+1.844669005, +1.844669005] CERTIFIED POSITIVE; in-regime True (max marg 0.31777); dichotomy True; 0s +[E] exact MU(22)_t181/16: aggregate in [+1.730725429, +1.730725429] CERTIFIED POSITIVE; in-regime True (max marg 0.28114); dichotomy True; 5s +[E] exact MU(32)_t181/16: aggregate in [+1.730725429, +1.730725429] CERTIFIED POSITIVE; in-regime True (max marg 0.28114); dichotomy True; 5s +[E] exact MU(32)_t4: aggregate in [+1.018477406, +1.018477406] CERTIFIED POSITIVE; in-regime True (max marg 0.27862); dichotomy True; 5s +[E] exact search_best_n8_t181/16: aggregate in [+0.166100824, +0.166100824] CERTIFIED POSITIVE; in-regime True (max marg 0.36657); dichotomy True; 0s +done in 181s; checkpoints in /home/user/math-lab/problems/union-closed/data diff --git a/problems/union-closed/explore/uc_or_agg_probe.py b/problems/union-closed/explore/uc_or_agg_probe.py new file mode 100644 index 0000000..2e50437 --- /dev/null +++ b/problems/union-closed/explore/uc_or_agg_probe.py @@ -0,0 +1,594 @@ +#!/usr/bin/env python3 +"""Attempt 014: large-n probe of the i-AGGREGATED odds-ratio control. + +Object under test (007 lead 1 / 013 claim 6 -- the restated Gap 1 after the +per-i averaged form M_i >= lambda died float-free in 007/013): + + A(mu, lambda) = sum_{i= 0 + for all mu, all lambda > 0, + +with M_i the well-posed history-averaged log2 odds ratio of 007 section 1 +(mass-weighted mean of log2 OR_i(a,b) over the nondegenerate history block +N_i x N_i of the tilt coupling pi_lambda(A,B) prop u(A)u(B)2^{lambda|A^B|}), +and w_i = defined history mass at coordinate i. The i = n term is excluded +(M_n = lambda identically, 005 Prop 1, so it contributes 0 anyway). + +KILL condition (exactly as certified in 013 part C, sign reversed): an +instance with A(mu, lambda) < 0 certified in exact rational arithmetic +(rational tilt t, conjecture reads sum_i sum_rows m*log2(OR/t) < 0), with +all elementwise marginals < 38271/100000 (the record regime), H(mu) > 0, +positive defined mass. Certified positives to date live at n <= 7 only +(+1.844669 on the 10-atom witness, 013 part C); 012's budget-census reversal +past n ~ 22 is the standing warning against small-n optimism -- hence this +probe BEFORE any proof effort. + +Parts: + A anchors: witness aggregate at n = 7 (target +1.8446); MU(n,1) exact + invariance at n = 20, 32 (padding coordinates carry w_i = 0); + crash/mirror embedded at n = 8..32 (<= 4-atom supports, aggregate >= 0 + is a THEOREM there via 007 Theorem B -- pure engine check); dichotomy + flags everywhere. + B constructive multi-unit witness ladder MU(n, r): r crash units + (per-unit light slice {1,c_j,n} + block-b responder {c_j,n-1}) sharing + the witness's two-block geometry, dilution auto-tuned into the record + regime; aggregate and per-i profile vs n at lambda in {2, 3.5, 5} -- + the margin-vs-n trend for the recorded killer genre. + C seeded adversarial hill-climbs per n in {8,...,32}: seeds = MU ladders, + witness-spread, sparsified Sawin geometric mixtures, delta_0+Bern + genre, random sparse supports; support mutations (bit toggles, atom + birth/death) + multiplicative weight noise; objective = aggregate + + penalties (marginal cap 0.375, defined mass, entropy); dynamic-range + clamp (1e-9 rel) and cell-floor diagnostics per 007's underflow guard. + Checkpointed per n. + D permutation attack on the best endpoints (coordinate order is part of + mu; the aggregate is NOT permutation-invariant): random + structured + coordinate permutations, hunting an order that flips the sign. + E exact rational certification (engine imported from 013's + uc_or_avg_skeptic: Fraction kernel + certified log2 enclosures) of the + extreme points: witness anchor at t = 181/16 vs 013's +1.844669, the + search minimum, and the largest-n ladder instances. + +Standard library only. Deterministic (fixed seeds, fixed step counts); a +wall-clock emergency abort exists but its firing is recorded in the +checkpoint. Runtime ~45-75 min single process. + +Usage: + python problems/union-closed/explore/uc_or_agg_probe.py [--fast] +Checkpoints: problems/union-closed/data/aggprobe_part[ABCDE]*.json + problems/union-closed/data/aggprobe_run.log (tee by hand) +""" + +from __future__ import annotations + +import json +import math +import random +import sys +import time +from fractions import Fraction +from pathlib import Path + +HERE = Path(__file__).resolve().parent +ROOT = HERE.parent +DATA = ROOT / "data" +sys.path.insert(0, str(HERE)) + +import uc_or_avg as ORA # 007's engine (float census) +import uc_or_avg_skeptic as ORS # 013's engine (exact rational) + +RECORD = 0.38271 +MARG_TARGET = 0.375 +CLAMP_REL = 1e-9 +T0 = time.monotonic() +FAST = "--fast" in sys.argv +# emergency abort only -- steps are fixed so the run is deterministic unless +# this fires (and then the checkpoint says so) +WALL_LIMIT = 300.0 if FAST else 4800.0 + + +def save(name: str, obj) -> None: + DATA.mkdir(exist_ok=True) + (DATA / name).write_text(json.dumps(obj, indent=1, sort_keys=True, + default=str) + "\n", encoding="utf-8") + + +def log(msg: str) -> None: + print(msg, flush=True) + + +# ---------------------------------------------------------------------- +# aggregate evaluation (float path; reuses 007's census verbatim) +# ---------------------------------------------------------------------- + +def agg_eval(u: dict[int, float], n: int, lam: float, want_cens=False): + """Returns dict with aggregate over i dict[int, float]: + """dynamic-range clamp: drop atoms below CLAMP_REL of the max weight + (honest support change, handled by the N_i bookkeeping).""" + mx = max(u.values()) + return {a: w for a, w in u.items() if w >= mx * CLAMP_REL} + + +# ---------------------------------------------------------------------- +# instance builders +# ---------------------------------------------------------------------- + +WIT = ORA.WITNESS_U # 007's 10-atom witness, n = 7, lambda = 3.5 + + +def mu_ladder(n: int, r: int, wdil: float = 20.0) -> dict[int, float]: + """MU(n, r): multi-unit witness. Coord 1 = block marker, coords 2-4 = + dilution singletons, response coords 5..4+r, shared futures n-1, n. + MU(7, 1) is exactly 007's witness. Requires 1 <= r <= n - 6.""" + assert 1 <= r <= n - 6 + b_hi, b_lo = 1 << (n - 1), 1 << (n - 2) # coords n, n-1 + u: dict[int, float] = { + 0b1 | b_lo: 3.802705744653898, # {1, n-1} + 0b1 | b_hi: 8.094727225702503, # {1, n} + 0b0: 0.2526660836353666, # {} + b_lo: 27.28641199464732, # {n-1} + b_hi: 12.529780963546393, # {n} + } + for j in range(r): + c = 1 << (4 + j) # response coord 5+j + u[0b1 | c | b_hi] = 0.004792968160237178 # light {1, c_j, n} + u[c | b_lo] = 0.3609234435898425 # {c_j, n-1} + for d in range(3): + u[1 << (1 + d)] = wdil # dilution {2},{3},{4} + return u + + +def mu_ladder_tuned(n: int, r: int, lam: float): + """Bisect the shared dilution weight so max marginal <= MARG_TARGET + (dilution coords' own marginals rise as others fall; a feasible window + usually exists). Returns (u, info).""" + lo, hi = 0.5, 400.0 + best = None + for _ in range(22): + w = math.sqrt(lo * hi) + u = mu_ladder(n, r, w) + ev = agg_eval(u, n, lam) + mm = ev["max_marginal"] + pi = ORA.tilt_of_potential(u, lam) + mu = ORA.mu_of(pi) + marg = ORA.marginal_elements(mu, n) + dil_m = max(marg[1:4]) + oth_m = max(m for j, m in enumerate(marg) if j not in (1, 2, 3)) + best = (u, {"wdil": w, "max_marginal": mm, "dil_marg": dil_m, + "other_marg": oth_m} | {k: ev[k] for k in + ("agg", "per_i_min", "argmin_i", + "H", "wsum")}) + if oth_m > MARG_TARGET and dil_m < MARG_TARGET: + lo = w + elif dil_m > MARG_TARGET: + hi = w + else: + break + return best + + +def witness_spread(n: int) -> dict[int, float]: + """007's witness embedded in [n] with its response coordinate moved to + the middle and futures at the top: coords (1..7) -> (1, 2, 3, 4, m, + n-1, n) with m = n//2. Padding coords start empty (w_i = 0); climb + mutations populate them.""" + m = n // 2 + perm = {0: 0, 1: 1, 2: 2, 3: 3, 4: m - 1, 5: n - 2, 6: n - 1} + u = {} + for a, w in WIT.items(): + b = 0 + for j in range(7): + if a >> j & 1: + b |= 1 << perm[j] + u[b] = w + return u + + +def sawin_sparse(n: int, k: int, rng) -> dict[int, float]: + """Sparsified Sawin-genre geometric mixture: sample k atoms from + sum_j (1-q) q^j Bern(p_j), p_j in a geometric ladder up to 0.92, and use + the (unnormalized) mixture density as the potential u (fixed-u trick: + any positive u is its own marginal's potential).""" + q = 0.55 + ps = [0.92 * (0.55 ** j) + 0.02 for j in range(6)] + u: dict[int, float] = {} + for _ in range(k * 3): + j = min(int(math.log(max(rng.random(), 1e-12)) / math.log(q)), 5) + p = ps[j] + a = 0 + for c in range(n): + if rng.random() < p: + a |= 1 << c + dens = sum((1 - q) * q ** jj * + (ps[jj] ** ORA.popcount(a)) * + ((1 - ps[jj]) ** (n - ORA.popcount(a))) + for jj in range(6)) + u[a] = dens + if len(u) >= k: + break + mx = max(u.values()) + return {a: w / mx for a, w in u.items()} + + +def d0bern_sparse(n: int, k: int, rng) -> dict[int, float]: + """delta_0 + Bern(1/2 + eps) genre, sparsified: heavy empty-set atom + plus k sampled Bernoulli atoms.""" + eps = 0.04 + u: dict[int, float] = {0: 1.0} + while len(u) < k + 1: + a = 0 + for c in range(n): + if rng.random() < 0.5 + eps: + a |= 1 << c + u[a] = 0.35 / k + return u + + +def rand_sparse(n: int, k: int, rng) -> dict[int, float]: + atoms = rng.sample(range(1 << n), k) if n <= 20 else \ + [rng.getrandbits(n) for _ in range(k)] + return {a: math.exp(rng.uniform(-1.5, 1.5)) for a in set(atoms)} + + +# ---------------------------------------------------------------------- +# hill-climb toward an aggregate violation +# ---------------------------------------------------------------------- + +def objective(ev): + if ev is None: + return float("inf") + return (ev["agg"] + + 25.0 * max(0.0, ev["max_marginal"] - MARG_TARGET) + + 10.0 * max(0.0, 0.05 - ev["wsum"]) + + 2.0 * max(0.0, 0.5 - ev["H"])) + + +def climb(u0: dict[int, float], n: int, lam: float, rng, steps: int, + k_max: int = 64): + u = clamp_u(dict(u0)) + ev = agg_eval(u, n, lam) + cur = objective(ev) + if not math.isfinite(cur): + return None, None + for step in range(steps): + if time.monotonic() - T0 > WALL_LIMIT: + break + sigma = 1.0 * (1 - step / steps) + 0.05 + cand = dict(u) + z = rng.random() + if z < 0.55: # weight jiggle + a = rng.choice(list(cand)) + cand[a] = max(cand[a], 1e-8) * math.exp(rng.gauss(0, sigma)) + elif z < 0.70: # light-slice push + a = min(cand, key=cand.get) + cand[a] = cand[a] * math.exp(rng.gauss(0, 2 * sigma)) + elif z < 0.85: # bit toggle (support move) + a = rng.choice(list(cand)) + b = a ^ (1 << rng.randrange(n)) + w = cand.pop(a) + cand[b] = cand.get(b, 0.0) + w + elif z < 0.95 and len(cand) < k_max: # atom birth + a = rng.choice(list(cand)) + b = a ^ (1 << rng.randrange(n)) + if rng.random() < 0.5: + b ^= 1 << rng.randrange(n) + if b not in cand: + cand[b] = math.exp(rng.uniform(-6, 0)) + elif len(cand) > 6: # atom death + a = min(cand, key=cand.get) + cand.pop(a) + cand = clamp_u(cand) + ev2 = agg_eval(cand, n, lam) + v2 = objective(ev2) + if v2 < cur: + u, ev, cur = cand, ev2, v2 + return u, ev + + +# ---------------------------------------------------------------------- +# part A -- anchors +# ---------------------------------------------------------------------- + +def part_a() -> dict: + out: dict = {} + ev = agg_eval(WIT, 7, 3.5, want_cens=True) + log(f"[A] witness n=7 lam=3.5: aggregate {ev['agg']:+.6f} " + f"(007/013: +1.8446 / +1.844669), per-i min {ev['per_i_min']:+.6f} " + f"at i={ev['argmin_i']}, max marginal {ev['max_marginal']:.4f}, " + f"dichotomy {ev['dichotomy_ok']}") + out["witness_n7"] = {k: v for k, v in ev.items() if k != "cens"} + + inv = {} + for n in (20, 32): + e2 = agg_eval(mu_ladder(n, 1), n, 3.5) + inv[n] = e2["agg"] + log(f"[A] MU({n},1) padding invariance: aggregate {e2['agg']:+.6f} " + f"(must equal witness {ev['agg']:+.6f}; diff " + f"{abs(e2['agg'] - ev['agg']):.2e})") + out["padding_invariance"] = {"witness": ev["agg"], "per_n": inv} + + thm_b = [] + for n in (8, 16, 24, 32): + for name, mu in (("crash", ORA.crash_mu(n)), ("mirror", ORA.mirror_mu(n))): + for lam in (0.5, 3.5): + u, _r = ORA.sinkhorn(mu, lam, tol=1e-12) + # sinkhorn returns pi; evaluate aggregate directly from pi + cens = ORA.census(u, n) + rows = [(c["defined_mass"], c["M_i"], c["i"]) for c in cens + if c["M_i"] is not None and c["i"] < n] + agg = (sum(w * (v - lam) for w, v, _i in rows) + / sum(w for w, _v, _i in rows)) + thm_b.append({"n": n, "family": name, "lambda": lam, + "agg": agg}) + worst = min(r["agg"] for r in thm_b) + log(f"[A] <=4-atom families (crash/mirror, n up to 32, Theorem-B-safe): " + f"{len(thm_b)} instances, min aggregate {worst:+.3e} (>= 0 expected)") + out["theorem_b_anchors"] = {"rows": thm_b, "min_agg": worst} + save("aggprobe_partA.json", out) + return out + + +# ---------------------------------------------------------------------- +# part B -- the ladder trend +# ---------------------------------------------------------------------- + +def part_b() -> dict: + rows = [] + ns = (7, 10, 13, 16, 19, 22, 25, 28, 32) if not FAST else (7, 10, 13) + for lam in (2.0, 3.5, 5.0): + for n in ns: + for r_mode, r in (("full", n - 6), ("fixed6", min(6, n - 6))): + if r < 1 or (r_mode == "fixed6" and n - 6 <= 6): + continue + u, info = mu_ladder_tuned(n, r, lam) + rows.append({"n": n, "r": r, "r_mode": r_mode, "lambda": lam} + | info) + for r_mode in ("full", "fixed6"): + tr = [r for r in rows if r["lambda"] == lam + and r["r_mode"] == r_mode] + if tr: + log(f"[B] MU ladder lam={lam} {r_mode}: agg vs n: " + + " ".join(f"{r['n']}:{r['agg']:+.4f}" for r in tr)) + log(f"[B] per-i min vs n: " + + " ".join(f"{r['n']}:{r['per_i_min']:+.4f}" for r in tr)) + log(f"[B] max marginal vs n: " + + " ".join(f"{r['n']}:{r['max_marginal']:.3f}" for r in tr)) + save("aggprobe_partB.json", {"rows": rows}) + return {"rows": rows} + + +# ---------------------------------------------------------------------- +# part C -- seeded adversarial search per n +# ---------------------------------------------------------------------- + +def part_c() -> dict: + ns = (8, 12, 16, 20, 24, 28, 32) if not FAST else (8, 12) + summary = [] + best_overall = None + for n in ns: + rng = random.Random(140000 + n) + steps = 500 if n <= 16 else (400 if n <= 24 else 300) + if FAST: + steps = 60 + seeds = [] + r_full = min(n - 6, 14) + seeds.append(("MU_full_3.5", mu_ladder(n, r_full), 3.5)) + seeds.append(("MU_full_2.0", mu_ladder(n, r_full), 2.0)) + if n - 6 >= 3: + seeds.append(("MU_r3_3.5", mu_ladder(n, 3), 3.5)) + seeds.append(("witness_spread_3.5", witness_spread(n), 3.5)) + seeds.append(("sawin_sparse_2.0", sawin_sparse(n, 18, rng), 2.0)) + seeds.append(("d0bern_3.5", d0bern_sparse(n, 15, rng), 3.5)) + seeds.append(("rand_2.0", rand_sparse(n, 12, rng), 2.0)) + seeds.append(("rand_5.0", rand_sparse(n, 10, rng), 5.0)) + endpoints = [] + best = None + for name, u0, lam in seeds: + t_start = time.monotonic() + u_end, ev = climb(u0, n, lam, rng, steps) + if ev is None: + endpoints.append({"seed": name, "lambda": lam, "agg": None}) + continue + rec = {"seed": name, "lambda": lam, "agg": ev["agg"], + "max_marginal": ev["max_marginal"], "H": ev["H"], + "wsum": ev["wsum"], "per_i_min": ev["per_i_min"], + "argmin_i": ev["argmin_i"], "k_atoms": ev["k_atoms"], + "secs": round(time.monotonic() - t_start, 1), + "steps": steps} + endpoints.append(rec) + in_reg = ev["max_marginal"] < RECORD + if best is None or (in_reg and ev["agg"] < best["agg"]): + best = rec | {"u": {format(a, "b").zfill(n): w + for a, w in u_end.items()}} + nviol = sum(1 for e in endpoints if e["agg"] is not None + and e["agg"] < -1e-9 and e["max_marginal"] < RECORD) + lows = sorted(e["agg"] for e in endpoints if e["agg"] is not None) + log(f"[C] n={n}: {len(endpoints)} climbs x {steps} steps, in-regime " + f"aggregate violations: {nviol}; three lowest endpoints: " + + " ".join(f"{v:+.5f}" for v in lows[:3]) + + f" (deepest per-i min {min(e['per_i_min'] for e in endpoints if e['agg'] is not None):+.4f})") + chk = {"n": n, "steps": steps, "endpoints": endpoints, + "n_in_regime_violations": nviol, "best": best, + "wall_abort": time.monotonic() - T0 > WALL_LIMIT} + save(f"aggprobe_partC_n{n}.json", chk) + summary.append({"n": n, "lowest": lows[0] if lows else None, + "n_violations": nviol}) + if best is not None and (best_overall is None + or best["agg"] < best_overall["agg"]): + best_overall = best | {"n": n} + if time.monotonic() - T0 > WALL_LIMIT: + log(f"[C] WALL LIMIT hit after n={n}; remaining n skipped") + break + save("aggprobe_partC_summary.json", {"summary": summary, + "best_overall": best_overall}) + return {"summary": summary, "best_overall": best_overall} + + +# ---------------------------------------------------------------------- +# part D -- permutation attack +# ---------------------------------------------------------------------- + +def apply_perm(u: dict[int, float], perm: list[int], n: int): + out = {} + for a, w in u.items(): + b = 0 + for j in range(n): + if a >> j & 1: + b |= 1 << perm[j] + out[b] = out.get(b, 0.0) + w + return out + + +def part_d(best_overall) -> dict: + rng = random.Random(140999) + targets = [("witness_n7", WIT, 7, 3.5)] + for n in ((13, 22, 32) if not FAST else (13,)): + if n - 6 >= 1: + u, _ = mu_ladder_tuned(n, min(n - 6, 14), 3.5) + targets.append((f"MU({n})", u, n, 3.5)) + if best_overall is not None and best_overall.get("u"): + n = best_overall["n"] + u = {int(k, 2): v for k, v in best_overall["u"].items()} + targets.append((f"search_best_n{n}", u, n, best_overall["lambda"])) + out_rows = [] + for name, u, n, lam in targets: + base = agg_eval(u, n, lam) + vals = [] + perms = [] + for _ in range(40 if not FAST else 6): + p = list(range(n)) + rng.shuffle(p) + perms.append(p) + perms.append(list(reversed(range(n)))) # full reversal + for p in perms: + ev = agg_eval(apply_perm(u, p, n), n, lam) + if ev is not None: + vals.append((ev["agg"], ev["max_marginal"])) + worst = min(v for v, _m in vals) + nviol = sum(1 for v, m in vals if v < -1e-9 and m < RECORD) + log(f"[D] permutation attack {name}: base {base['agg']:+.4f}, " + f"{len(vals)} orders, min {worst:+.4f}, in-regime violations " + f"{nviol}") + out_rows.append({"target": name, "n": n, "lambda": lam, + "base_agg": base["agg"], "min_over_perms": worst, + "n_violations": nviol, "n_perms": len(vals)}) + save("aggprobe_partD.json", {"rows": out_rows}) + return {"rows": out_rows} + + +# ---------------------------------------------------------------------- +# part E -- exact rational certification of the extreme points +# ---------------------------------------------------------------------- + +def exact_aggregate(u_float: dict[int, float], n: int, t: Fraction): + """Certified enclosure of the aggregate at rational tilt t for the exact + rational measure whose potential is the float-rationalization of u. + Reuses 013's exact_census / log2 enclosures verbatim.""" + u = {a: Fraction(w) for a, w in u_float.items()} + cens = ORS.exact_census(u, t, n) + lo_sum = hi_sum = Fraction(0) + wtot = Fraction(0) + dich = True + for c in cens: + if c["i"] >= n or not c["rows"]: + continue + dich = dich and c["dich_ok"] + for (_a, _b, m, orr) in c["rows"]: + l, h = ORS.log2_enclosure(orr / t) + lo_sum += m * l + hi_sum += m * h + wtot += c["def_mass"] + _mu, marg = ORS.exact_marginals(u, t, n) + return {"agg_lo": lo_sum / wtot, "agg_hi": hi_sum / wtot, + "max_marg": max(marg), "in_regime": max(marg) < ORS.RECORD, + "dich_ok": dich} + + +def dec_bound(x: Fraction, digits: int = 21, up: bool = False) -> str: + """Directed-rounded decimal string for a certified bound (avoids + serializing astronomically long exact fractions).""" + s = 10 ** digits + num, den = x.numerator * s, x.denominator + q = -((-num) // den) if up else num // den + sign = "-" if q < 0 else "+" + q = abs(q) + return f"{sign}{q // s}.{q % s:0{digits}d}" + + +def part_e(best_overall) -> dict: + t_wit = Fraction(181, 16) # lambda = log2(181/16) ~ 3.50022 + out: dict = {} + jobs = [("witness_n7_t181/16", WIT, 7, t_wit)] + for n in ((22, 32) if not FAST else (13,)): + if n - 6 >= 1: + u, _ = mu_ladder_tuned(n, min(n - 6, 14), 3.5) + jobs.append((f"MU({n})_t181/16", u, n, t_wit)) + jobs.append(("MU(32)_t4", mu_ladder_tuned(32, 14, 2.0)[0], 32, + Fraction(4)) if not FAST else + ("MU(13)_t4", mu_ladder_tuned(13, 7, 2.0)[0], 13, + Fraction(4))) + if best_overall is not None and best_overall.get("u"): + n = best_overall["n"] + u = {int(k, 2): v for k, v in best_overall["u"].items()} + lam = best_overall["lambda"] + t = {2.0: Fraction(4), 3.5: t_wit, 5.0: Fraction(32), + 0.5: Fraction(181, 128)}.get(lam, Fraction(2.0 ** lam)) + jobs.append((f"search_best_n{n}_t{t}", u, n, t)) + rows = [] + for name, u, n, t in jobs: + t_start = time.monotonic() + r = exact_aggregate(u, n, t) + lo, hi = float(r["agg_lo"]), float(r["agg_hi"]) + verdict = ("CERTIFIED NEGATIVE (KILL)" if r["agg_hi"] < 0 else + "CERTIFIED POSITIVE" if r["agg_lo"] > 0 else "STRADDLES 0") + log(f"[E] exact {name}: aggregate in [{lo:+.9f}, {hi:+.9f}] " + f"{verdict}; in-regime {r['in_regime']} " + f"(max marg {float(r['max_marg']):.5f}); dichotomy {r['dich_ok']}; " + f"{time.monotonic() - t_start:.0f}s") + rows.append({"name": name, "n": n, "t": str(t), + "agg_lo": dec_bound(r["agg_lo"], up=False), + "agg_hi": dec_bound(r["agg_hi"], up=True), + "agg_lo_float": lo, "agg_hi_float": hi, + "in_regime": r["in_regime"], + "max_marg_float": float(r["max_marg"]), + "dich_ok": r["dich_ok"], "verdict": verdict}) + out["rows"] = rows + save("aggprobe_partE.json", out) + return out + + +def main() -> None: + part_a() + part_b() + c = part_c() + part_d(c["best_overall"]) + part_e(c["best_overall"]) + log(f"done in {time.monotonic() - T0:.0f}s; checkpoints in {DATA}") + + +if __name__ == "__main__": + main() diff --git a/problems/union-closed/explore/uc_or_agg_probe2.py b/problems/union-closed/explore/uc_or_agg_probe2.py new file mode 100644 index 0000000..7035d71 --- /dev/null +++ b/problems/union-closed/explore/uc_or_agg_probe2.py @@ -0,0 +1,631 @@ +#!/usr/bin/env python3 +"""Attempt 014, deep phase: push the aggregated-control probe past n = 32. + +Round 1 (uc_or_agg_probe.py part B) found the multi-unit witness ladder +MU(n, n-6) has aggregate decaying like a - b log n over n = 7..32 (best +2-parameter fit; max residual ~0.05), which EXTRAPOLATES to a sign change at +n ~ 78 (lambda 5), ~140 (lambda 3.5), ~320 (lambda 2). Extrapolation is not +evidence (012's lesson cuts both ways), so this phase measures instead: + + P1 ladder extension: MU(n, n-6), dilution tuned on marginals only, + n up to 128, lambda in {2, 3.5} (+5 at spot points). Sign check. + P2 unit-symmetric ladder optimization: the ladder is determined by 8 + shared weights (A1, A2, light, B0, B1, B2, resp, dil); climb THETA at + moderate n (in-regime penalty), then evaluate the optimized theta at + large n. This is the orbit-symmetrization idea in search form: the + symmetric parametrization keeps the eval cheap and the search + low-dimensional at n far beyond round 1's free-support climbs. + P3 deep annealed free-support climbs at n = 24, 32, seeded from round + 1's best endpoints (checkpoints on disk) + fresh ladders; 2500 steps, + stiffer marginal penalty, occasional uphill acceptance. + P4 lambda profiles of the extreme instances (violations must not hide + between grid points). + P5 exact rational certification (013's engine) of the decisive points: + any float violation found, else the tightest in-regime margins at the + largest n reached, in tidy-rational form (limit_denominator) so the + certification stays tractable at large n. + +Standard library only. Deterministic (fixed seeds). Runtime ~30-60 min. + +Usage: + python problems/union-closed/explore/uc_or_agg_probe2.py +Checkpoints: problems/union-closed/data/aggprobe2_partP[12345]*.json + problems/union-closed/data/aggprobe2_run.log (tee by hand) +""" + +from __future__ import annotations + +import json +import math +import random +import sys +import time +from fractions import Fraction +from pathlib import Path + +HERE = Path(__file__).resolve().parent +ROOT = HERE.parent +DATA = ROOT / "data" +sys.path.insert(0, str(HERE)) + +import uc_or_avg as ORA +import uc_or_avg_skeptic as ORS +from uc_or_agg_probe import (agg_eval, clamp_u, mu_ladder, objective, + exact_aggregate, dec_bound, RECORD, MARG_TARGET) + +T0 = time.monotonic() + + +def save(name: str, obj) -> None: + DATA.mkdir(exist_ok=True) + (DATA / name).write_text(json.dumps(obj, indent=1, sort_keys=True, + default=str) + "\n", encoding="utf-8") + + +def log(msg: str) -> None: + print(msg, flush=True) + + +# ---------------------------------------------------------------------- +# fast marginal-only evaluation (dilution tuning at large n must not pay +# for the census) +# ---------------------------------------------------------------------- + +def marginals_of_u(u: dict[int, float], n: int, lam: float): + supp = list(u) + z = 0.0 + mu = {} + for a in supp: + row = sum(u[a] * u[b] * 2.0 ** (lam * ORA.popcount(a & b)) + for b in supp) + mu[a] = row + z += row + marg = [0.0] * n + for a, w in mu.items(): + for j in range(n): + if a >> j & 1: + marg[j] += w / z + return marg + + +THETA_WITNESS = { + "A1": 3.802705744653898, "A2": 8.094727225702503, + "light": 0.004792968160237178, "B0": 0.2526660836353666, + "B1": 27.28641199464732, "B2": 12.529780963546393, + "resp": 0.3609234435898425, "dil": 20.0, +} + + +def mu_ladder_theta(n: int, r: int, th: dict[str, float]) -> dict[int, float]: + assert 1 <= r <= n - 6 + b_hi, b_lo = 1 << (n - 1), 1 << (n - 2) + u = {0b1 | b_lo: th["A1"], 0b1 | b_hi: th["A2"], 0b0: th["B0"], + b_lo: th["B1"], b_hi: th["B2"]} + for j in range(r): + c = 1 << (4 + j) + u[0b1 | c | b_hi] = th["light"] + u[c | b_lo] = th["resp"] + for d in range(3): + u[1 << (1 + d)] = th["dil"] + return u + + +def tune_dil(n: int, r: int, lam: float, th=None): + """Bisect the dilution weight on marginals only.""" + th = dict(th or THETA_WITNESS) + lo, hi = 0.5, 2000.0 + for _ in range(18): + w = math.sqrt(lo * hi) + th["dil"] = w + marg = marginals_of_u(mu_ladder_theta(n, r, th), n, lam) + dil_m = max(marg[1:4]) + oth_m = max(m for j, m in enumerate(marg) if j not in (1, 2, 3)) + if oth_m > MARG_TARGET and dil_m < MARG_TARGET: + lo = w + elif dil_m > MARG_TARGET: + hi = w + else: + break + return th + + +# ---------------------------------------------------------------------- +# P1 -- ladder extension +# ---------------------------------------------------------------------- + +def part_p1() -> dict: + rows = [] + for lam, ns in ((2.0, (40, 48, 64, 80, 96, 128)), + (3.5, (40, 48, 64, 80, 96, 128)), + (5.0, (48, 96))): + for n in ns: + r = n - 6 + th = tune_dil(n, r, lam) + t0 = time.monotonic() + ev = agg_eval(mu_ladder_theta(n, r, th), n, lam) + rows.append({"n": n, "r": r, "lambda": lam, "wdil": th["dil"], + "agg": ev["agg"], "max_marginal": ev["max_marginal"], + "per_i_min": ev["per_i_min"], "H": ev["H"], + "argmin_i": ev["argmin_i"], + "secs": round(time.monotonic() - t0, 1)}) + log(f"[P1] MU({n},{r}) lam={lam}: agg {ev['agg']:+.4f}, " + f"max marg {ev['max_marginal']:.3f}, per-i min " + f"{ev['per_i_min']:+.4f} ({rows[-1]['secs']}s)") + save("aggprobe2_partP1.json", {"rows": rows}) + return {"rows": rows} + + +# ---------------------------------------------------------------------- +# P2 -- unit-symmetric theta optimization +# ---------------------------------------------------------------------- + +def theta_objective(th, n, r, lam): + ev = agg_eval(clamp_u(mu_ladder_theta(n, r, th)), n, lam) + if ev is None: + return float("inf"), None + pen = (100.0 * max(0.0, ev["max_marginal"] - MARG_TARGET) + + 10.0 * max(0.0, 0.05 - ev["wsum"]) + + 2.0 * max(0.0, 0.5 - ev["H"])) + return ev["agg"] + pen, ev + + +def part_p2() -> dict: + out_rows = [] + best_by_lam = {} + n_opt, r_opt = 48, 42 + for lam in (2.0, 3.5): + rng = random.Random(150000 + int(lam * 10)) + th = dict(tune_dil(n_opt, r_opt, lam)) + cur, ev = theta_objective(th, n_opt, r_opt, lam) + for step in range(260): + sigma = 0.8 * (1 - step / 260) + 0.05 + cand = dict(th) + k = rng.choice(list(cand)) + cand[k] = max(cand[k], 1e-8) * math.exp(rng.gauss(0, sigma)) + v2, ev2 = theta_objective(cand, n_opt, r_opt, lam) + if v2 < cur: + th, cur, ev = cand, v2, ev2 + log(f"[P2] theta-opt at n={n_opt} lam={lam}: agg {ev['agg']:+.4f} " + f"(mm {ev['max_marginal']:.3f}), theta=" + + " ".join(f"{k}:{v:.4g}" for k, v in sorted(th.items()))) + best_by_lam[lam] = th + out_rows.append({"stage": "opt", "n": n_opt, "lambda": lam, + "theta": th, "agg": ev["agg"], + "max_marginal": ev["max_marginal"], + "per_i_min": ev["per_i_min"]}) + for n in (64, 96, 128): + r = n - 6 + th_n = tune_dil(n, r, lam, th) + ev_n = agg_eval(mu_ladder_theta(n, r, th_n), n, lam) + log(f"[P2] theta({lam}) at n={n}: agg {ev_n['agg']:+.4f}, " + f"mm {ev_n['max_marginal']:.3f}, per-i min " + f"{ev_n['per_i_min']:+.4f}") + out_rows.append({"stage": "transfer", "n": n, "lambda": lam, + "theta": th_n, "agg": ev_n["agg"], + "max_marginal": ev_n["max_marginal"], + "per_i_min": ev_n["per_i_min"]}) + save("aggprobe2_partP2.json", {"rows": out_rows}) + return {"rows": out_rows, "best_by_lam": best_by_lam} + + +# ---------------------------------------------------------------------- +# P3 -- deep annealed free-support climbs at n = 24, 32 +# ---------------------------------------------------------------------- + +def anneal(u0, n, lam, rng, steps=2500, k_max=72): + u = clamp_u(dict(u0)) + ev = agg_eval(u, n, lam) + + def obj(e): + if e is None: + return float("inf") + return (e["agg"] + 100.0 * max(0.0, e["max_marginal"] - MARG_TARGET) + + 10.0 * max(0.0, 0.05 - e["wsum"]) + + 2.0 * max(0.0, 0.5 - e["H"])) + + cur = obj(ev) + if not math.isfinite(cur): + return None, None + best_u, best_ev, best_v = dict(u), ev, cur + for step in range(steps): + temp = 0.25 * (1 - step / steps) + 0.002 + sigma = 1.0 * (1 - step / steps) + 0.05 + cand = dict(u) + z = rng.random() + if z < 0.5: + a = rng.choice(list(cand)) + cand[a] = max(cand[a], 1e-8) * math.exp(rng.gauss(0, sigma)) + elif z < 0.65: + a = min(cand, key=cand.get) + cand[a] = cand[a] * math.exp(rng.gauss(0, 2 * sigma)) + elif z < 0.82: + a = rng.choice(list(cand)) + b = a ^ (1 << rng.randrange(n)) + w = cand.pop(a) + cand[b] = cand.get(b, 0.0) + w + elif z < 0.94 and len(cand) < k_max: + a = rng.choice(list(cand)) + b = a ^ (1 << rng.randrange(n)) + if rng.random() < 0.5: + b ^= 1 << rng.randrange(n) + if b not in cand: + cand[b] = math.exp(rng.uniform(-6, 0)) + elif len(cand) > 6: + cand.pop(min(cand, key=cand.get)) + cand = clamp_u(cand) + ev2 = agg_eval(cand, n, lam) + v2 = obj(ev2) + if v2 < cur or rng.random() < math.exp(-(v2 - cur) / temp): + u, cur, ev = cand, v2, ev2 + if cur < best_v: + best_u, best_ev, best_v = dict(u), ev, cur + return best_u, best_ev + + +def load_round1_seed(n: int): + f = DATA / f"aggprobe_partC_n{n}.json" + if not f.exists(): + return None + d = json.loads(f.read_text()) + b = d.get("best") + if not b or not b.get("u"): + return None + return {int(k, 2): v for k, v in b["u"].items()}, b["lambda"], b["seed"] + + +def part_p3() -> dict: + endpoints = [] + best = None + for n in (24, 32): + rng = random.Random(151000 + n) + seeds = [] + r1 = load_round1_seed(n) + if r1: + seeds.append((f"round1_best({r1[2]})", r1[0], r1[1])) + seeds.append(("MU_full_2.0", mu_ladder(n, min(n - 6, 18)), 2.0)) + seeds.append(("MU_full_3.5", mu_ladder(n, min(n - 6, 18)), 3.5)) + for name, u0, lam in seeds: + t0 = time.monotonic() + u_end, ev = anneal(u0, n, lam, rng) + if ev is None: + continue + rec = {"n": n, "seed": name, "lambda": lam, "agg": ev["agg"], + "max_marginal": ev["max_marginal"], + "per_i_min": ev["per_i_min"], "argmin_i": ev["argmin_i"], + "H": ev["H"], "wsum": ev["wsum"], + "k_atoms": ev["k_atoms"], + "secs": round(time.monotonic() - t0, 1)} + endpoints.append(rec) + log(f"[P3] n={n} {name} lam={lam}: agg {ev['agg']:+.5f}, " + f"mm {ev['max_marginal']:.4f}, per-i min " + f"{ev['per_i_min']:+.4f} ({rec['secs']}s)") + in_reg = ev["max_marginal"] < RECORD + if in_reg and (best is None or ev["agg"] < best["agg"]): + best = rec | {"u": {format(a, "b").zfill(n): w + for a, w in u_end.items()}} + nviol = sum(1 for e in endpoints + if e["agg"] < -1e-9 and e["max_marginal"] < RECORD) + log(f"[P3] deep climbs: {len(endpoints)} endpoints, in-regime " + f"violations {nviol}, best in-regime " + f"{best['agg'] if best else None}") + save("aggprobe2_partP3.json", {"endpoints": endpoints, "best": best, + "n_violations": nviol}) + return {"endpoints": endpoints, "best": best} + + +# ---------------------------------------------------------------------- +# P4 -- lambda profiles of the extreme instances +# ---------------------------------------------------------------------- + +def part_p4(p2, p3) -> dict: + grid = (0.25, 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 5.0, 6.0) + out = {} + jobs = [] + th2 = p2["best_by_lam"].get(2.0) + if th2: + jobs.append(("theta2_n96_ladder", None, 96, th2)) + if p3["best"]: + b = p3["best"] + jobs.append((f"p3_best_n{b['n']}", + {int(k, 2): v for k, v in b["u"].items()}, b["n"], None)) + for name, u_fixed, n, th in jobs: + prof = [] + for lam in grid: + if th is not None: + th_n = tune_dil(n, n - 6, lam, th) + u = mu_ladder_theta(n, n - 6, th_n) + else: + u = u_fixed + ev = agg_eval(u, n, lam) + prof.append({"lambda": lam, "agg": None if ev is None + else ev["agg"], + "max_marginal": None if ev is None + else ev["max_marginal"]}) + vals = [(p["lambda"], p["agg"]) for p in prof if p["agg"] is not None] + log(f"[P4] lambda profile {name}: " + + " ".join(f"{l}:{a:+.3f}" for l, a in vals)) + out[name] = prof + save("aggprobe2_partP4.json", out) + return out + + +# ---------------------------------------------------------------------- +# P5 -- exact certification of the decisive points +# ---------------------------------------------------------------------- + +def tidy_u(u: dict[int, float], den: int = 10 ** 4) -> dict[int, Fraction]: + """limit_denominator can round a tiny positive float to exactly 0; a + zero-weight atom left in the support is a phantom atom that breaks the + degeneracy dichotomy's support bookkeeping, so drop zeros AFTER + rationalizing (caught by the dich_ok=False flag on the first P5 run's + P3-best certificate; the kill certificates never had weights small + enough to be affected).""" + out = {a: Fraction(w).limit_denominator(den) for a, w in u.items() + if w > 0} + return {a: w for a, w in out.items() if w > 0} + + +def part_p5(p1, p2, p3) -> dict: + t_by_lam = {2.0: Fraction(4), 3.5: Fraction(181, 16), + 5.0: Fraction(32)} + jobs = [] + # the float-minimum ladder points at the largest n we can afford exactly + lam_min_row = min(p1["rows"], key=lambda r: r["agg"]) + jobs.append((f"P1_min_MU({lam_min_row['n']})_lam{lam_min_row['lambda']}", + mu_ladder_theta(lam_min_row["n"], lam_min_row["r"], + tune_dil(lam_min_row["n"], lam_min_row["r"], + lam_min_row["lambda"])), + lam_min_row["n"], t_by_lam[lam_min_row["lambda"]])) + tr = [r for r in p2["rows"] if r["stage"] == "transfer"] + if tr: + r_min = min(tr, key=lambda r: r["agg"]) + jobs.append((f"P2_min_theta_n{r_min['n']}_lam{r_min['lambda']}", + mu_ladder_theta(r_min["n"], r_min["n"] - 6, + r_min["theta"]), + r_min["n"], t_by_lam[r_min["lambda"]])) + if p3["best"]: + b = p3["best"] + jobs.append((f"P3_best_n{b['n']}_lam{b['lambda']}", + {int(k, 2): v for k, v in b["u"].items()}, + b["n"], t_by_lam.get(b["lambda"], + Fraction(2.0 ** b["lambda"])))) + rows = [] + for name, u_f, n, t in jobs: + u = tidy_u(u_f if isinstance(next(iter(u_f.values())), float) + else u_f) + t0 = time.monotonic() + r = exact_aggregate({a: w for a, w in u.items()}, n, t) + verdict = ("CERTIFIED NEGATIVE (KILL)" if r["agg_hi"] < 0 else + "CERTIFIED POSITIVE" if r["agg_lo"] > 0 else + "STRADDLES 0") + log(f"[P5] exact tidy-rational {name}: aggregate in " + f"[{float(r['agg_lo']):+.9f}, {float(r['agg_hi']):+.9f}] " + f"{verdict}; in-regime {r['in_regime']} (max marg " + f"{float(r['max_marg']):.5f}); dichotomy {r['dich_ok']}; " + f"{time.monotonic() - t0:.0f}s") + rows.append({"name": name, "n": n, "t": str(t), + "agg_lo": dec_bound(r["agg_lo"], up=False), + "agg_hi": dec_bound(r["agg_hi"], up=True), + "agg_lo_float": float(r["agg_lo"]), + "agg_hi_float": float(r["agg_hi"]), + "in_regime": r["in_regime"], + "max_marg_float": float(r["max_marg"]), + "dich_ok": r["dich_ok"], "verdict": verdict, + "u_tidy": {format(a, "b").zfill(n): str(w) + for a, w in u.items()}}) + save("aggprobe2_partP5.json", {"rows": rows}) + return {"rows": rows} + + +# ---------------------------------------------------------------------- +# P6 (--sweep) -- adversarial dilution-weight sweep: the tuned-family trend +# of P1/part B picks SOME feasible dilution weight per n, but the adversary +# owns mu, so the honest per-n ladder margin is the MINIMUM over the +# dilution weight subject to in-regime marginals. +# ---------------------------------------------------------------------- + +def part_p6() -> dict: + grid = [0.5 * (2000 / 0.5) ** (k / 15) for k in range(16)] + rows = [] + for lam in (2.0, 3.5): + for n in (10, 16, 22, 28, 32, 40, 48, 64, 80, 96, 128): + r = n - 6 + best = None + th = dict(THETA_WITNESS) + for w in grid: + th["dil"] = w + ev = agg_eval(mu_ladder_theta(n, r, th), n, lam) + if ev is None or ev["max_marginal"] >= RECORD: + continue + if best is None or ev["agg"] < best["agg"]: + best = {"wdil": w, "agg": ev["agg"], + "max_marginal": ev["max_marginal"], + "per_i_min": ev["per_i_min"]} + rows.append({"n": n, "r": r, "lambda": lam} | (best or {})) + log(f"[P6] MU({n},{r}) lam={lam}: min-over-wdil agg " + f"{best['agg']:+.4f} at wdil {best['wdil']:.1f} " + f"(mm {best['max_marginal']:.3f})" if best else + f"[P6] MU({n},{r}) lam={lam}: no in-regime point") + tr = [r for r in rows if r["lambda"] == lam and "agg" in r] + log(f"[P6] lam={lam} adversarial-dil trend: " + + " ".join(f"{r['n']}:{r['agg']:+.4f}" for r in tr)) + save("aggprobe2_partP6.json", {"rows": rows}) + return {"rows": rows} + + +# ---------------------------------------------------------------------- +# P7 (--kill) -- drill into a float-negative aggregate found by P2: n-scan +# of the optimized theta, perturbation stability, and exact rational +# certification at exactly-rational tilt (t = 4, lambda = 2 exactly; both +# the float-exact dyadic weights and a tidy limit_denominator version). +# ---------------------------------------------------------------------- + +def part_p7() -> dict: + d = json.loads((DATA / "aggprobe2_partP2.json").read_text()) + cands = [r for r in d["rows"] if r["lambda"] == 2.0] + th = min(cands, key=lambda r: r["agg"])["theta"] + lam = 2.0 + out: dict = {"theta": th} + + # (a) n-scan: where does the optimized ladder cross zero? + scan = [] + for n in (48, 56, 64, 72, 80, 88, 96, 104, 112, 120, 128, 144, 160): + th_n = tune_dil(n, n - 6, lam, th) + ev = agg_eval(mu_ladder_theta(n, n - 6, th_n), n, lam) + scan.append({"n": n, "agg": ev["agg"], "wdil": th_n["dil"], + "max_marginal": ev["max_marginal"], + "per_i_min": ev["per_i_min"], "H": ev["H"], + "wsum": ev["wsum"]}) + log(f"[P7a] n={n}: agg {ev['agg']:+.6f}, mm " + f"{ev['max_marginal']:.3f}, H {ev['H']:.2f}") + out["n_scan"] = scan + neg = [s for s in scan if s["agg"] < -1e-9 and s["max_marginal"] < RECORD] + tgt = min(neg, key=lambda s: s["agg"]) if neg else \ + min(scan, key=lambda s: s["agg"]) + n_t = tgt["n"] + th_t = tune_dil(n_t, n_t - 6, lam, th) + u_t = mu_ladder_theta(n_t, n_t - 6, th_t) + log(f"[P7] drill target n={n_t}: float agg {tgt['agg']:+.6f}") + + # (b) perturbation stability: 20 x 3% multiplicative noise on ALL weights + rng = random.Random(160001) + pert = [] + for _ in range(20): + up = {a: w * math.exp(rng.gauss(0, 0.03)) for a, w in u_t.items()} + ev = agg_eval(up, n_t, lam) + pert.append({"agg": ev["agg"], "max_marginal": ev["max_marginal"]}) + nneg = sum(1 for p in pert if p["agg"] < -1e-9 + and p["max_marginal"] < RECORD) + log(f"[P7b] perturbation: {nneg}/20 remain in-regime negative " + f"(range {min(p['agg'] for p in pert):+.6f} .. " + f"{max(p['agg'] for p in pert):+.6f})") + out["perturbation"] = {"rows": pert, "n_negative": nneg} + + # (c) cell floor at the deficit coordinates (underflow guard) + pi = ORA.tilt_of_potential(u_t, lam) + ev_t = agg_eval(u_t, n_t, lam, want_cens=True) + out["cell_floor_at_argmin"] = ev_t["cell_floor_at_argmin"] + log(f"[P7c] cell floor at argmin i={ev_t['argmin_i']}: " + f"{ev_t['cell_floor_at_argmin']:.2e}") + + # (d) exact certification at t = 4 (lambda = 2 exactly) + rows = [] + for tag, u_ex in (("float_exact_dyadic", {a: Fraction(w) + for a, w in u_t.items()}), + ("tidy_1e4", tidy_u(u_t))): + t0 = time.monotonic() + r = exact_aggregate(u_ex, n_t, Fraction(4)) + verdict = ("CERTIFIED NEGATIVE (KILL)" if r["agg_hi"] < 0 else + "CERTIFIED POSITIVE" if r["agg_lo"] > 0 else + "STRADDLES 0") + log(f"[P7d] exact {tag} n={n_t} t=4: aggregate in " + f"[{float(r['agg_lo']):+.9f}, {float(r['agg_hi']):+.9f}] " + f"{verdict}; in-regime {r['in_regime']} (max marg " + f"{float(r['max_marg']):.5f}); dichotomy {r['dich_ok']}; " + f"{time.monotonic() - t0:.0f}s") + rows.append({"tag": tag, "n": n_t, "t": "4", + "agg_lo": dec_bound(r["agg_lo"], up=False), + "agg_hi": dec_bound(r["agg_hi"], up=True), + "agg_lo_float": float(r["agg_lo"]), + "agg_hi_float": float(r["agg_hi"]), + "in_regime": r["in_regime"], + "max_marg_float": float(r["max_marg"]), + "dich_ok": r["dich_ok"], "verdict": verdict}) + out["exact"] = rows + out["u_target"] = {format(a, "b").zfill(n_t): w for a, w in u_t.items()} + save("aggprobe2_partP7.json", out) + return out + + +# ---------------------------------------------------------------------- +# P8 (--cert96) -- exact certification of the theta*(2.0) ladder at the +# crossing point n = 96, float-exact dyadic weights, t = 4. (First run as +# an inline snippet; kept here so every certificate has a reproducible +# command.) Writes aggprobe2_n96cert.json. +# ---------------------------------------------------------------------- + +def part_p8() -> dict: + d = json.loads((DATA / "aggprobe2_partP2.json").read_text()) + row = [r for r in d["rows"] if r["stage"] == "transfer" + and r["lambda"] == 2.0 and r["n"] == 96][0] + u = mu_ladder_theta(96, 90, row["theta"]) + t0 = time.monotonic() + r = exact_aggregate({a: Fraction(w) for a, w in u.items()}, 96, + Fraction(4)) + verdict = ("CERTIFIED NEGATIVE (KILL)" if r["agg_hi"] < 0 else + "CERTIFIED POSITIVE" if r["agg_lo"] > 0 else "STRADDLES 0") + out = {"name": "theta2_n96_float_exact_dyadic_t4", "n": 96, "t": "4", + "agg_lo": dec_bound(r["agg_lo"], up=False), + "agg_hi": dec_bound(r["agg_hi"], up=True), + "agg_lo_float": float(r["agg_lo"]), + "agg_hi_float": float(r["agg_hi"]), + "in_regime": r["in_regime"], + "max_marg_float": float(r["max_marg"]), + "dich_ok": r["dich_ok"], "verdict": verdict, + "secs": round(time.monotonic() - t0, 1)} + log(f"[P8] exact theta*(2.0) n=96 t=4: aggregate in " + f"[{out['agg_lo_float']:+.9f}, {out['agg_hi_float']:+.9f}] " + f"{verdict}; in-regime {out['in_regime']} " + f"(max marg {out['max_marg_float']:.5f}); dichotomy " + f"{out['dich_ok']}; {out['secs']}s") + save("aggprobe2_n96cert.json", out) + return out + + +# ---------------------------------------------------------------------- +# P9 (--certp3) -- re-certification of the P3-best n = 32 endpoint with +# zero-dropped tidy weights (the first P5 pass left a phantom zero-weight +# atom after limit_denominator, tripping the exact dichotomy flag). +# Writes aggprobe2_p3best_cert.json. +# ---------------------------------------------------------------------- + +def part_p9() -> dict: + d = json.loads((DATA / "aggprobe2_partP3.json").read_text()) + b = d["best"] + u = tidy_u({int(k, 2): v for k, v in b["u"].items()}) + r = exact_aggregate(dict(u), b["n"], Fraction(4)) + verdict = ("CERTIFIED NEGATIVE (KILL)" if r["agg_hi"] < 0 else + "CERTIFIED POSITIVE" if r["agg_lo"] > 0 else "STRADDLES 0") + out = {"name": f"P3_best_n{b['n']}_lam2.0_tidy_zero_dropped", + "n": b["n"], "t": "4", + "agg_lo": dec_bound(r["agg_lo"], up=False), + "agg_hi": dec_bound(r["agg_hi"], up=True), + "agg_lo_float": float(r["agg_lo"]), + "agg_hi_float": float(r["agg_hi"]), + "in_regime": r["in_regime"], + "max_marg_float": float(r["max_marg"]), + "dich_ok": r["dich_ok"], "verdict": verdict} + log(f"[P9] exact P3-best n={b['n']} t=4 (zero-dropped tidy): " + f"[{out['agg_lo_float']:+.9f}, {out['agg_hi_float']:+.9f}] " + f"{verdict}; dichotomy {out['dich_ok']}") + save("aggprobe2_p3best_cert.json", out) + return out + + +def main() -> None: + if "--sweep" in sys.argv: + part_p6() + log(f"done in {time.monotonic() - T0:.0f}s") + return + if "--cert96" in sys.argv: + part_p8() + log(f"done in {time.monotonic() - T0:.0f}s") + return + if "--certp3" in sys.argv: + part_p9() + log(f"done in {time.monotonic() - T0:.0f}s") + return + if "--kill" in sys.argv: + part_p7() + log(f"done in {time.monotonic() - T0:.0f}s") + return + p1 = part_p1() + p2 = part_p2() + p3 = part_p3() + part_p4(p2, p3) + part_p5(p1, p2, p3) + log(f"done in {time.monotonic() - T0:.0f}s") + + +if __name__ == "__main__": + main() diff --git a/problems/union-closed/prior-art.json b/problems/union-closed/prior-art.json index ebd532d..b85eda7 100644 --- a/problems/union-closed/prior-art.json +++ b/problems/union-closed/prior-art.json @@ -439,6 +439,42 @@ "minor ratio 6.68" ], "note": "Decisive check: 007's confessed vulnerability (both engines share IEEE floats; its first witnesses were shared-underflow artifacts) is closed by re-verifying the witness in exact rational arithmetic with directed-rounding log2 enclosures. At t = 4 and t = 16 the refutation of M_i >= lambda is a fully rational statement with lambda = 2, 4 exactly. New strengthening beyond 007: the violation does not hinge on the conditioning-out convention (degenerate histories scored at exactly lambda still give M_5 - lambda = -0.067, certified). Nuance: exact certificates at lambda = 1, 2, 3 are for rationalized-refit measures within 9.2e-16 of the witness mu, not the identical rational mu. Theorem C's smoothness step is analogous to but not literally discharged by 008's P6-prime (P6-prime is d/d-mu at fixed lambda; Theorem C needs d/d-lambda at fixed mu)." + }, + { + "id": "014", + "file": "attempts/014-aggregated-control-probe-large-n.md", + "date": "2026-08-04", + "mode": "informed", + "status": "REFUTED", + "mechanism": [ + "coupling", + "overlap-tilt", + "sinkhorn", + "plackett", + "gram-mass-pairing", + "exact-rational-arithmetic" + ], + "one_line": "Large-n probe of the restated Gap 1 kills it: unit-replicating the 10-atom witness genre (r crash units sharing one block marker) with re-optimized shared weights drives the i-aggregated control negative past n ~ 90, certified float-free at t = 4 (n = 96 and 160, marginals <= 0.309); the n <= 32 space stays positive everywhere searched.", + "leak_terms": [ + "unit replication", + "multi-unit witness ladder", + "MU(n,r)", + "theta-star ladder", + "aggregate crossing", + "O(1) surplus corridor", + "0.024226", + "-0.000759865", + "uc_or_agg_probe", + "aggprobe" + ], + "range": "Free-support seeded climbs n <= 32 (58 climbs round 1, deep annealed round 2) plus permutation orbits: 0 violations; unit-symmetric ladder evaluated to n = 160 (raw witness weights: positive through n = 128, min +0.157; optimized theta*: float crossing in (88, 96], -0.0242 at n = 160, 20/20 perturbations stay negative); exact rational certificates at t = 181/16 and t = 4, enclosure widths < 1e-18", + "refuted_by": null, + "gaps": [ + "margin-modulated-or-control", + "mutual-information-tax", + "recipe-totality" + ], + "note": "Settles queue item 1's probe-first branch negatively: the aggregated control is certified positive at n <= 32 on everything searched, so proof effort there was exactly the small-n trap 012 warned about. The kill is convention-robust (scoring degenerate histories at lambda only grows the denominator, never the numerator). Route stays LIVE; Gap 1 falls to the margin-modulated candidate (007 lead 2), which this record's ladder family must now be run against." } ] -} +} \ No newline at end of file From 425c4015b46007b17313cfcd2a46c64002aa7ca3 Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 10:57:05 +0000 Subject: [PATCH 10/14] union-closed 015: skeptic review of 014 - the kill stands, two corrections Fully independent confirmation: quantity re-derived from 007/013 prose and re-implemented from scratch (scaled-integer arithmetic, own log2 enclosure via interval squaring + directed dyadic rounding), hitting the 013 anchor digit-for-digit and re-deriving all three kill certificates to 18+ digits (n=96 enclosure strictly negative, width 1.3e-20). Family rebuilt atom-for-atom from prose; marginals and dichotomy certified as integer inequalities; positive controls confirmed and the surviving raw-weight ladder upgraded to certified +0.157. Corrections: C1 cosmetic (n=128 has 252 atoms, not 253); C2 real - the violation lives at lambda in [2,2.5], NOT in the 009/011 workable window (~0.05 at n=96), so the recorded forall-lambda gap is validly closed but the lambda <~ c/n restricted variant is untouched and stays live alongside the margin-modulated control. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- .../attempts/015-skeptic-review-of-014.md | 331 ++++++ .../union-closed/data/aggsk15_results.json | 146 +++ .../union-closed/data/aggsk15_selftest.json | 87 ++ .../union-closed/explore/uc_or_agg_skeptic.py | 486 +++++++++ problems/union-closed/prior-art.json | 997 +++++++++--------- 5 files changed, 1568 insertions(+), 479 deletions(-) create mode 100644 problems/union-closed/attempts/015-skeptic-review-of-014.md create mode 100644 problems/union-closed/data/aggsk15_results.json create mode 100644 problems/union-closed/data/aggsk15_selftest.json create mode 100644 problems/union-closed/explore/uc_or_agg_skeptic.py diff --git a/problems/union-closed/attempts/015-skeptic-review-of-014.md b/problems/union-closed/attempts/015-skeptic-review-of-014.md new file mode 100644 index 0000000..b9b21c0 --- /dev/null +++ b/problems/union-closed/attempts/015-skeptic-review-of-014.md @@ -0,0 +1,331 @@ +# 015 — Skeptic review of 014 (aggregated control killed at large n): independent exact recomputation + +- **Problem:** union-closed, `problems/union-closed/PROBLEM.md` +- **Date:** 2026-08-04 +- **Mode:** informed +- **Type:** adversarial verification of `014-aggregated-control-probe-large-n.md` + (default stance: refute the refutation — a wrong kill would close a live + proof branch incorrectly). 014's certificates reuse 013's exact engine and + 013's log₂ enclosures; the decisive check here is therefore a **from-scratch + third implementation of the aggregate**, derived from 007 §1's prose (not + transcribed from any engine), in scaled-integer exact arithmetic with a + **different certified log₂ enclosure** (interval squaring with directed + dyadic rounding + the exact series ln 2 = Σ 1/(k·2^k), not 013's atanh + series). Per the verification standard (2026-07-31), exact rational + arithmetic IS the skeptic pass for claims decidable over ℚ; this review + additionally removes the shared-enclosure-code and shared-census-code risks + 013's Residual-risk section named. +- **Outcome in one line:** **014 holds — the kill stands.** All three kill + certificates (n = 96, 128, 160 at t = 4) re-derive from scratch with + matching digits and certified negative sign; the positive controls + (+1.844669005 witness anchor, P3-best n = 32, the raw-weight ladder at + n = 96) re-certify positive; the construction, marginals, dichotomy, and + convention-robustness claims all check exactly. Two corrections, neither + load-bearing for the kill: a wrong atom count for the n = 128 certificate + (252, not 253), and P4's claim that the violating λ-window sits "exactly + where 009/011's λ-window law puts the workable range" — it does not: at + n = 96 that law puts the assembly's workable λ at ≈ 0.052, and the + violation lives at λ ∈ [2, 2.5], far outside it. The formal ∀λ statement + is dead; the λ-restricted restatement (014's own lead 4) is more alive + than the record's P4 framing suggests. +- **Tools:** `explore/uc_or_agg_skeptic.py` (written here; stdlib only; + imports NOTHING from `uc_or_avg*.py` / `uc_or_agg_probe*.py`; instance + parameters are read as data from 014's checkpoints, and the MU(n,r) + builder is re-implemented from 014 §1's prose then cross-checked + atom-by-atom against the probe builder in a separate snippet; + deterministic, no randomness; full run with `--n128 --n160` ≈ 15 s). + Command reproducing every number below: + `python problems/union-closed/explore/uc_or_agg_skeptic.py --n128 --n160` + Checkpoints: `data/aggsk15_selftest.json`, `data/aggsk15_results.json`. +- **Sources:** 007/013 (definition of M_i, w_i, the aggregate, the anchors), + 014 (the record under attack and its checkpoints), 005 (Prop 1 anchor), + 009/011 (λ-window law, for the design audit), STATUS.md verification + standard. No external fetches. + +Notation as in 007/013: tilt coupling π ∝ u(A)u(B)·t^{|A∩B|}, t = 2^λ; +M_i the mass-weighted mean of log₂ OR_i over N_i²; w_i the defined history +mass; A(μ,λ) = Σ_{i 0 + ("for all μ and all i", 005 lead 1 as restated by 007; the control is a + lemma about arbitrary tilt marginals, applied downstream to + union-closed-supported measures). MU's support is not union-closed — + and neither was 007's 10-atom witness that 013 certified. Checked and + reported by the engine (`union_closed_support: false` for every + instance including the witness) so no future reader mistakes this for + an unexamined gap. + +### 3. The kill certificates (claim 3) — CONFIRMED from scratch; the kill is real + +My engine's certified enclosures (interval-squaring log₂, width ≤ 1.3e-20 +per instance; every OR/t handled as an exact rational; dedup by value — +277 distinct log arguments at n = 96): + + n = 96, 188 atoms, t = 4: A ∈ [−0.000759865185844166092, + −0.000759865185844166079] KILL (1.1s) + n = 128, 252 atoms, t = 4: A ∈ [−0.015405726918 ± 7e-21] KILL (3.2s) + n = 160, 316 atoms, t = 4: A ∈ [−0.024225909315 ± 7e-21] KILL (7.2s) + +Each interval **contains 014's certified enclosure** (n = 96: +[−...087, −...086]; n = 128: −0.0154057269180687; n = 160: +−0.02422590931474056) and is strictly negative, with in-regime marginals +and exact dichotomy. Two structurally different enclosure methods (013's +atanh-series with 1e-50 directed rationalization vs my dyadic +interval-squaring with the ln 2 series bound) and two independently +written censuses agree to 18+ digits. The sign at n = 96 is a theorem +about the stated 188 rational weights. λ = 2 exactly (t = 4): no +irrational number appears in the statement. + +### 4. The positive controls (claim 4) — CONFIRMED; the probe is not broken-negative + +- Witness anchor: above (+1.844669005341). +- **P3-best n = 32** (tidy, zero-dropped, 36 atoms): my engine certifies + +0.297679573398 at t = 4, dichotomy exact — matching 014's re-cert and + confirming the phantom-zero-atom postmortem (the `limit_denominator` + zero-drop is the correct fix; with zeros dropped the dichotomy flag is + clean in my fully independent bookkeeping too). +- **"Surviving positive space" is real:** the RAW-witness-weight ladder at + n = 96 with P6's adversarial dilution weight (wdil = 41.6935…) certifies + **+0.157131854664** at t = 4 — my certificate upgrades 014's float + +0.157132 (P6 was float-only) to a certified positive. The kill is + genuinely a re-weighting result, not pure replication, exactly as the + record's "calibrates how much of the kill is re-weighting" lead assumes. +- Round-1 exact positives cross-checked against `aggprobe_partE.json` + (witness +1.8446690053408297, MU(22)=MU(32) +1.7307254294897931, + MU(32)_t4 +1.0184774056855224, n = 8 endpoint +0.1661008241): the + record quotes them faithfully. + +### 5. Robustness claims (claim 5) — CONFIRMED as scoped + +- **Convention robustness:** re-derived the one-line proof. Under 013's + alternative bookkeeping (every degenerate positive-mass history scored + at exactly λ), each such history contributes m·(λ−λ) = 0 to the + numerator, and the per-i denominator grows from w_i to the full history + mass; so A_alt = N/((n−1)·Z) with the SAME numerator N. Sign invariance + is rigorous, and computed: A_alt at n = 96 ∈ [−0.000383136592916716685, + −0.000383136592916716678], certified negative (128: −0.007751; + 160: −0.012168). No bookkeeping convention rescues the statement on + these μ (only a convention awarding degenerate histories MORE than λ + could, which 013 already ruled indefensible). +- **Perturbation claim, scope pinned:** what was perturbed is the 316 + potential weights (3% multiplicative noise, 20 trials, n = 160), + evaluated in FLOAT (`aggprobe2_partP7.json` rows, range −0.0253..−0.0208, + all in-regime). This is EVIDENCE that the violation is an open set, not + part of the certificate — correctly so, since the certified statements + are about the exact stated weights and need no stability. No issue; a + reader should just not cite the 20/20 as certified. + +### 6. Design audit (claim 6) — kill valid for the stated gap; one framing corrected + +- The recorded gap (`aggregated-or-control`, 007 lead 1 = 013 claim 6 = + queue item 1) is the ∀μ ∀λ>0 statement; one certified (μ, λ) pair + refutes it, exactly as 007/013's per-i kill worked. The kill closes + queue item 1's "if it survives n ≳ 20, attack the proof" branch as + claimed: do not invest in the Gram/Frobenius proof of the unrestricted + aggregate. +- But see **C2**: the violating λ-window and the assembly's workable + window (λ_max ≈ 4.847/(n−3)) are disjoint at every violating n, so the + practical bite on the route is narrower than P4's framing suggests, and + lead 4 (λ-restricted control) is the natural next Gap-1 candidate + alongside the margin-modulated one. The Outcome/Not-claimed section + itself is accurate (it claims no ∀λ violation and flags the window); + the record's calibration survives except for the one P4 sentence. +- The float-crossing claim "(88, 96]" matches P7's n-scan data (+0.00456 + at n = 88, −0.00076 at 96) — float EVIDENCE, correctly labeled. +- Search-completeness claims (n ≤ 32 clean, 58+6 climbs, permutation + orbits) were spot-audited against checkpoints for consistency but NOT + re-run — they are scoped as EVIDENCE in 014 and carry no weight in the + kill; the certified n ≤ 32 positives (witness, P3-best, ladder points) + are what backs "a fixed-small-n control is untouched", and those I + re-certified. + +### 7. Enclosure smell test (claim 7) — the sign is 16 orders above the noise floor + +My log₂ enclosure self-test (`aggsk15_selftest.json`): exact on powers of +2 (width 0), correct on log₂ 3, log₂ 10, log₂(181/16), (7/5)^10, +1 + 1e-40 (width ≤ 1.5e-20 each), additivity-consistent, monotone. The +n = 96 aggregate enclosure width is 1.3e-20 against a certified value of +−7.6e-4: the sign has ~16 orders of margin. Underflow/cell-floor +pathologies are structurally impossible in the scaled-integer census +(every kernel cell is a positive integer; the 007-genre IEEE artifact +cannot occur). The dynamic-range worry from 007's detour does not apply +to the certificates at all — it applies only to the float searches that +found the instances, and their endpoints were then certified exactly. + +## Verdict + +| # | 014 claim | Verdict | +|---|-----------|---------| +| 1 | A(μ,λ) is 013's certified aggregate (007 §1 M_i, w_i weights, i < n) | **CONFIRMED** — from-scratch third implementation matches the +1.844669005 anchor and every certificate; i = n exclusion verified as an exact identity (Prop 1), inclusion variant still negative | +| 2 | n = 96/128/160 instances valid: MU prose = data, marginals < 0.38271, dichotomy, H | **CONFIRMED** (atom-for-atom builder match; marginals certified as integer inequalities, max 0.309; dichotomy exact; union-closure correctly not a hypothesis — the statement quantifies over all μ) | +| 3 | Certified kills: −0.000759865…, −0.0154057…, −0.0242259… at t = 4 | **CONFIRMED — the kill stands.** Independent census + independent log₂ enclosure re-derive all three signs with matching digits (widths ≤ 1.3e-20) | +| 4 | Positive controls (+1.844669, P3-best +0.29768, raw ladder positive) | **CONFIRMED**; raw-witness ladder at n = 96 upgraded from float to certified +0.157131854664 | +| 5 | Convention robustness; perturbation stability | **CONFIRMED** (one-line proof re-derived; A_alt certified negative at all three n; perturbation claim is float EVIDENCE, correctly scoped) | +| 6 | Ledger consequence: `aggregated-or-control` (∀λ) dies; branch closed | **CONFIRMED with C2**: the kill is valid for the stated gap, but the violating λ-window is disjoint from the assembly's workable window at every violating n — the λ-restricted restatement (lead 4) is untouched by this family and joins the margin-modulated control as a live Gap-1 candidate | +| 7 | "253 atoms" at n = 128 | **REFUTED (cosmetic)** — 252 atoms (C1) | + +**Net assessment:** 014's headline — the i-aggregated odds-ratio control +(the restated Gap 1) is REFUTED, including in the record-relevant marginal +regime, by the θ*-optimized replicated-witness ladder at n ≥ 96, certified +float-free at λ = 2 exactly — should be treated as **VERIFIED**. The +certificates are now backed by two independently written exact censuses +and two structurally different certified log₂ enclosures, closing the +shared-code residual risks 013 recorded. The record's self-calibration was +good: the parts it flagged for attack first (the exact certificates, the +definition match, the anchors) all survived; the one genuine correction is +in the downstream framing (C2), not the mathematics. + +## Residual risk + +- **Shared-definition risk, further reduced but not zero:** all five + engines formalize "conditional OR of the tilt coupling given histories" + from 003/005's definitions. My implementation was written from 007 §1's + prose, not from code, and hits the forced anchors (M_n = λ exactly, + padding invariance exactly) — but the prose itself is the shared + ancestor. A conceptual error in the *definition* of the route's Gap 1 + (as opposed to its computation) would pass every check here; 013's + mitigation (006 S1's slice/inner-product derivation) still carries that + weight. +- **Search-completeness (n ≤ 32 positive space)** was audited for + checkpoint consistency, not re-run. It is EVIDENCE in 014 and remains + EVIDENCE. +- **Float-level side claims** (crossing in (88, 96], perturbation 20/20, + P4 window edges, P6 sweep minima other than the one certified here) + were reconciled against checkpoints but not certified; only the n = 96 + raw-ladder point was upgraded to certified. +- My log₂ enclosure's correctness argument (interval squaring + crude + endpoint bounds divided by 2^64, ln 2 series with geometric tail) is + ~40 lines and self-tested, but has itself not been reviewed by anyone + else. It agrees with 013's independent enclosure on every instance both + have touched, which is the practical mitigation. + +## References + +- Reviewed record: `problems/union-closed/attempts/014-aggregated-control-probe-large-n.md`; + its tools `explore/uc_or_agg_probe.py`, `explore/uc_or_agg_probe2.py`; + checkpoints `data/aggprobe2_partP[1-7].json`, `data/aggprobe2_n96cert.json`, + `data/aggprobe2_p3best_cert.json`, `data/aggprobe_part[ABCDE]*.json`, logs + `data/aggprobe2_*.log`. +- Context: `attempts/007-averaged-or-control.md` (§1 definition, §5 + aggregate, witness); `attempts/013-skeptic-review-of-007.md` (part C + anchor, exact standard, alternative bookkeeping, residual risks); + `attempts/005-odds-ratio-control-refuted.md` (Prop 1, "for all μ" + quantifier); `attempts/009/011` (λ-window law λ_max ≈ 4.847/(n−3), 011's + measured λ_max table); `attempts/012` (large-n lesson, cited by 014). +- This review's tool/data: `explore/uc_or_agg_skeptic.py`; + `data/aggsk15_selftest.json`, `data/aggsk15_results.json`. +- No external sources; the ln 2 series Σ 1/(k·2^k) and the bounds + 1 − 1/v ≤ ln v ≤ v − 1 are standard calculus, re-derived in the tool's + docstrings. diff --git a/problems/union-closed/data/aggsk15_results.json b/problems/union-closed/data/aggsk15_results.json new file mode 100644 index 0000000..e29f840 --- /dev/null +++ b/problems/union-closed/data/aggsk15_results.json @@ -0,0 +1,146 @@ +{ + "p3best_n32_tidy_t4": { + "H_bits": 2.8359480623665654, + "W_over_Z_float": 10.357116816662238, + "agg_hi_float": 0.29767957339818424, + "agg_lo_float": 0.29767957339818424, + "alt_hi_float": 0.09945490695545726, + "alt_lo_float": 0.09945490695545726, + "argmax_marg_coord": 32, + "dich_ok": true, + "in_regime": true, + "incl_hi_float": 0.2967422648406048, + "incl_lo_float": 0.2967422648406048, + "k_atoms": 36, + "max_marg_float": 0.344552407031527, + "n": 32, + "n_log_terms": 48, + "n_rows": 66, + "prop1_ok": true, + "secs": 0.0, + "t": "4", + "union_closed_support": false, + "verdict": "CERTIFIED POSITIVE", + "width_float": 1.2444887665871741e-20 + }, + "raw_ladder_n96_t4": { + "H_bits": 3.918406685922122, + "W_over_Z_float": 20.389096098391168, + "agg_hi_float": 0.1571318546635414, + "agg_lo_float": 0.1571318546635414, + "alt_hi_float": 0.03372396299845663, + "alt_lo_float": 0.03372396299845663, + "argmax_marg_coord": 95, + "dich_ok": true, + "in_regime": true, + "incl_hi_float": 0.15706631632152956, + "incl_lo_float": 0.15706631632152956, + "k_atoms": 188, + "max_marg_float": 0.34713087978123874, + "n": 96, + "n_log_terms": 277, + "n_rows": 368, + "prop1_ok": true, + "secs": 1.0, + "t": "4", + "union_closed_support": false, + "verdict": "CERTIFIED POSITIVE", + "width_float": 1.3253801033997966e-20 + }, + "theta2_n128_tidy_t4": { + "H_bits": 2.996499640505702, + "W_over_Z_float": 63.89287117164365, + "agg_hi_float": -0.0154057269180687, + "agg_lo_float": -0.0154057269180687, + "alt_hi_float": -0.007750520671509341, + "alt_lo_float": -0.007750520671509341, + "argmax_marg_coord": 127, + "dich_ok": true, + "in_regime": true, + "incl_hi_float": -0.015396162134341626, + "incl_lo_float": -0.015396162134341626, + "k_atoms": 252, + "max_marg_float": 0.3079282992952311, + "n": 128, + "n_log_terms": 373, + "n_rows": 496, + "prop1_ok": true, + "secs": 3.2, + "t": "4", + "union_closed_support": false, + "verdict": "CERTIFIED NEGATIVE (KILL)", + "width_float": 1.3258863844548188e-20 + }, + "theta2_n160_dyadic_t4": { + "H_bits": 3.026418663312094, + "W_over_Z_float": 79.86098958491169, + "agg_hi_float": -0.02422590931474056, + "agg_lo_float": -0.02422590931474056, + "alt_hi_float": -0.012167956550122712, + "alt_lo_float": -0.012167956550122712, + "argmax_marg_coord": 159, + "dich_ok": true, + "in_regime": true, + "incl_hi_float": -0.024213931087201664, + "incl_lo_float": -0.024213931087201664, + "k_atoms": 316, + "max_marg_float": 0.3071897109038987, + "n": 160, + "n_log_terms": 469, + "n_rows": 624, + "prop1_ok": true, + "secs": 7.2, + "t": "4", + "union_closed_support": false, + "verdict": "CERTIFIED NEGATIVE (KILL)", + "width_float": 1.2531624005727306e-20 + }, + "theta2_n96_t4": { + "H_bits": 2.9666023130156622, + "W_over_Z_float": 47.9005710554459, + "agg_hi_float": -0.0007598651858441661, + "agg_lo_float": -0.0007598651858441661, + "alt_hi_float": -0.00038313659291671666, + "alt_lo_float": -0.00038313659291671666, + "argmax_marg_coord": 95, + "dich_ok": true, + "in_regime": true, + "incl_hi_float": -0.0007592330636543233, + "incl_lo_float": -0.0007592330636543233, + "k_atoms": 188, + "max_marg_float": 0.30866648372421185, + "n": 96, + "n_log_terms": 277, + "n_rows": 368, + "prop1_ok": true, + "secs": 1.1, + "t": "4", + "union_closed_support": false, + "verdict": "CERTIFIED NEGATIVE (KILL)", + "width_float": 1.2407262580845524e-20 + }, + "witness_t181_16": { + "H_bits": 2.721778724157644, + "W_over_Z_float": 3.5789449961099953, + "agg_hi_float": 1.8446690053408297, + "agg_lo_float": 1.8446690053408297, + "alt_hi_float": 1.1003281510239606, + "alt_lo_float": 1.1003281510239606, + "argmax_marg_coord": 6, + "dich_ok": true, + "in_regime": true, + "incl_hi_float": 1.8263212036479564, + "incl_lo_float": 1.8263212036479564, + "k_atoms": 10, + "max_marg_float": 0.31777018538674157, + "n": 7, + "n_log_terms": 10, + "n_rows": 12, + "prop1_ok": true, + "secs": 0.0, + "t": "181/16", + "union_closed_support": false, + "verdict": "CERTIFIED POSITIVE", + "width_float": 1.2193622675835958e-20 + } +} diff --git a/problems/union-closed/data/aggsk15_selftest.json b/problems/union-closed/data/aggsk15_selftest.json new file mode 100644 index 0000000..7afe7bc --- /dev/null +++ b/problems/union-closed/data/aggsk15_selftest.json @@ -0,0 +1,87 @@ +{ + "additivity_ok": true, + "cases": [ + { + "hi": 0.0, + "lo": 0.0, + "ok": true, + "q": "1", + "ref": 0.0, + "width": 0.0 + }, + { + "hi": 1.0, + "lo": 1.0, + "ok": true, + "q": "2", + "ref": 1.0, + "width": 0.0 + }, + { + "hi": 10.0, + "lo": 10.0, + "ok": true, + "q": "1024", + "ref": 10.0, + "width": 0.0 + }, + { + "hi": -3.0, + "lo": -3.0, + "ok": true, + "q": "1/8", + "ref": -3.0, + "width": 0.0 + }, + { + "hi": 1.584962500721156, + "lo": 1.584962500721156, + "ok": true, + "q": "3", + "ref": 1.584962500721156, + "width": 1.4917148882739233e-20 + }, + { + "hi": 3.321928094887362, + "lo": 3.321928094887362, + "ok": true, + "q": "10", + "ref": 3.321928094887362, + "width": 7.723357048073797e-22 + }, + { + "hi": -1.584962500721156, + "lo": -1.584962500721156, + "ok": true, + "q": "1/3", + "ref": -1.584962500721156, + "width": 5.309071319394233e-21 + }, + { + "hi": 3.499845887083205, + "lo": 3.499845887083205, + "ok": true, + "q": "181/16", + "ref": 3.499845887083205, + "width": 1.4348700234729205e-21 + }, + { + "hi": 4.854268271702417, + "lo": 4.854268271702417, + "ok": true, + "q": "282475249/9765625", + "ref": 4.854268271702416, + "width": 2.1370975324547594e-21 + }, + { + "hi": 1.4426950408889633e-40, + "lo": 1.4426950408889633e-40, + "ok": true, + "q": "10000000000000000000000000000000000000001/100000000000000000", + "ref": 0.0, + "width": 4.6010887675702125e-58 + } + ], + "ok": true, + "pow2_exact": true +} diff --git a/problems/union-closed/explore/uc_or_agg_skeptic.py b/problems/union-closed/explore/uc_or_agg_skeptic.py new file mode 100644 index 0000000..4cc44a9 --- /dev/null +++ b/problems/union-closed/explore/uc_or_agg_skeptic.py @@ -0,0 +1,486 @@ +#!/usr/bin/env python3 +"""Attempt 015: skeptic re-computation of 014's aggregated-control kill. + +INDEPENDENCE CONTRACT. This file imports nothing from uc_or_avg.py, +uc_or_avg_skeptic.py, uc_or_agg_probe.py or uc_or_agg_probe2.py. The +quantity is re-derived from the PROSE of 007 section 1 and 007 section 5 / +013 part C: + + tilt coupling pi(A,B) prop u(A)u(B) t^{|A&B|}, t = 2^lambda, both + marginals mu (any positive u is its own marginal's potential, 005); + for coordinate i (1..n; coordinate j is bit j-1), histories are the + (i-1)-bit prefixes; N_i = prefixes of supp(mu) with BOTH continuations + reachable; for (a,b) in N_i^2 the conditional 2x2 table over the + response bits has four positive cells and OR_i(a,b) is its odds ratio; + m_ab = pi-mass of the history pair; M_i = sum m_ab log2 OR / sum m_ab; + w_i = sum_{N_i^2} m_ab (defined history mass); + + A(mu, lambda) = sum_{i=1} 1/(k 2^k) with its elementary geometric tail bound. + +All couplings, tables, odds ratios and marginals are computed in scaled +INTEGER arithmetic (every input weight is rational; multiply through by the +lcm of denominators and by den(t)^n), so no rounding exists anywhere except +inside the log2 enclosure, which is directed. + +Standard library only. Deterministic. Instances are read as DATA from +014's checkpoints (theta parameters / atom lists); the family construction +MU(n, r) is re-implemented from the prose of 014 section 1 and +cross-checked atom-by-atom in scripts/ -- not imported. + +Usage: + python problems/union-closed/explore/uc_or_agg_skeptic.py [--quick] + [--n128] [--n160] +Checkpoints: problems/union-closed/data/aggsk15_*.json +""" + +from __future__ import annotations + +import json +import math +import sys +import time +from fractions import Fraction +from pathlib import Path + +ROOT = Path(__file__).resolve().parent.parent +DATA = ROOT / "data" + +RECORD_NUM, RECORD_DEN = 38271, 100000 # threshold 0.38271 (Liu) +F0 = Fraction(0) + +# ====================================================================== +# certified log2 enclosure -- interval squaring with directed rounding +# ====================================================================== + +# ln 2 = sum_{k>=1} 1/(k 2^k); partial sum is a strict lower bound and the +# tail after K terms is < sum_{k>K} 1/((K+1) 2^k) = 1/((K+1) 2^K). +_LN2_TERMS = 64 +LN2_LO = sum(Fraction(1, k * (1 << k)) for k in range(1, _LN2_TERMS + 1)) +LN2_HI = LN2_LO + Fraction(1, (_LN2_TERMS + 1) * (1 << _LN2_TERMS)) +assert LN2_LO < LN2_HI + + +def log2_enclosure(q: Fraction, K: int = 64, prec: int = 192 + ) -> tuple[Fraction, Fraction]: + """[lo, hi] with lo <= log2(q) <= hi, q > 0 rational. + + Method: write q = 2^e * x, x in [1, 2). Repeat K times: square the + dyadic interval enclosing x (lower endpoint rounded down, upper rounded + up, at 2^-prec resolution) and renormalize by powers of two, tracking + the accumulated exponent S = sum_k s_k 2^{K-k}. Then + log2 x = (S + log2 x_K) / 2^K + and x_K lies in a known tiny interval well inside (1/4, 4), where the + elementary bounds 1 - 1/v <= ln v <= v - 1 (all v > 0) and the LN2 + bounds above give a crude but certified enclosure of log2 x_K whose + error is divided by 2^K. Total width ~ 2^-60 or better. + """ + num, den = q.numerator, q.denominator + if num <= 0: + raise ValueError("log2 of nonpositive") + if num & (num - 1) == 0 and den & (den - 1) == 0: + e = (num.bit_length() - 1) - (den.bit_length() - 1) + return Fraction(e), Fraction(e) + e = num.bit_length() - den.bit_length() + if e >= 0: + nn, dd = num, den << e + else: + nn, dd = num << (-e), den + if nn < dd: # x < 1: double it + nn <<= 1 + e -= 1 + # now dd <= nn < 2 dd, i.e. x = nn/dd in [1, 2) + scale = 1 << prec + two_scale = scale << 1 + xlo = (nn << prec) // dd # floor + xhi = -((-(nn << prec)) // dd) # ceil + S = 0 + for _ in range(K): + xlo = (xlo * xlo) >> prec # floor(x^2 * scale) + xhi = -((-(xhi * xhi)) >> prec) # ceil + s = 0 + while xlo >= two_scale: # whole interval >= 2 + xlo >>= 1 + xhi = -((-xhi) >> 1) + s += 1 + while xhi < scale: # whole interval < 1 + xlo <<= 1 + xhi <<= 1 + s -= 1 + S = (S << 1) + s + vlo = Fraction(xlo, scale) + vhi = Fraction(xhi, scale) + assert 0 < vlo <= vhi + # certified crude bounds on log2 v over v in [vlo, vhi] + ln_lo = 1 - Fraction(1) / vlo # <= ln vlo + ln_hi = vhi - 1 # >= ln vhi + l2_lo = ln_lo / (LN2_HI if ln_lo >= 0 else LN2_LO) + l2_hi = ln_hi / (LN2_LO if ln_hi >= 0 else LN2_HI) + pw = 1 << K + return e + Fraction(S + l2_lo, 1) / pw, e + Fraction(S + l2_hi, 1) / pw + + +def _selftest_log2() -> dict: + """Certified-enclosure self-test on known values (attack item g).""" + out = {"cases": [], "ok": True} + cases = [ + (Fraction(1), 0.0), (Fraction(2), 1.0), (Fraction(1024), 10.0), + (Fraction(1, 8), -3.0), (Fraction(3), math.log2(3)), + (Fraction(10), math.log2(10)), (Fraction(1, 3), -math.log2(3)), + (Fraction(181, 16), math.log2(181 / 16)), + (Fraction(7, 5) ** 10, 10 * math.log2(1.4)), + (Fraction(10 ** 40 + 1, 10 ** 40), math.log2(1 + 1e-40)), + ] + for q, ref in cases: + lo, hi = log2_enclosure(q) + w = float(hi - lo) + ok = lo <= hi and w < 1e-17 and lo - 1e-12 <= ref <= hi + 1e-12 + out["cases"].append({"q": str(q)[:60], "ref": ref, "lo": float(lo), + "hi": float(hi), "width": w, "ok": ok}) + out["ok"] = out["ok"] and ok + # additivity consistency: log2(ab) inside sum of enclosures + a, b = Fraction(355, 113), Fraction(1234567, 7654321) + la, ha = log2_enclosure(a) + lb, hb = log2_enclosure(b) + lab, hab = log2_enclosure(a * b) + add_ok = la + lb <= hab and lab <= ha + hb + out["additivity_ok"] = bool(add_ok) + out["ok"] = out["ok"] and add_ok + # exactness on powers of two + lo, hi = log2_enclosure(Fraction(1, 2 ** 100)) + out["pow2_exact"] = bool(lo == hi == -100) + out["ok"] = out["ok"] and out["pow2_exact"] + return out + + +# ====================================================================== +# exact aggregate (scaled-integer census) +# ====================================================================== + +def exact_aggregate(u: dict[int, Fraction], n: int, t: Fraction, + progress: bool = False) -> dict: + """Certified enclosure of A(mu, lambda) at rational tilt t for the + measure whose tilt potential is u (lambda = log2 t). Everything up to + the log2 calls is exact integer arithmetic.""" + t_start = time.monotonic() + supp = sorted(u) + assert all(w > 0 for w in u.values()), "zero/negative weight atom" + # scale weights to integers + L = 1 + for w in u.values(): + L = L * w.denominator // math.gcd(L, w.denominator) + U = {a: int(u[a] * L) for a in supp} + tn, td = t.numerator, t.denominator + # kernel k(A,B) = U_A U_B tn^p td^(n-p), p = |A & B| (global scale + # L^2 td^n; constant across pairs and coordinates, cancels everywhere) + pw_tn = [1] * (n + 1) + pw_td = [1] * (n + 1) + for j in range(1, n + 1): + pw_tn[j] = pw_tn[j - 1] * tn + pw_td[j] = pw_td[j - 1] * td + m_atoms = len(supp) + kern_rows: list[list[int]] = [] + for A in supp: + UA = U[A] + row = [] + for B in supp: + p = bin(A & B).count("1") + row.append(UA * U[B] * pw_tn[p] * pw_td[n - p]) + kern_rows.append(row) + Z = sum(sum(r) for r in kern_rows) # total coupling mass, scaled + + # exact elementwise marginals of mu: mu(A) = sum_B k(A,B) + mu_row = [sum(r) for r in kern_rows] + marg_num = [0] * n + for idx, A in enumerate(supp): + v = mu_row[idx] + for j in range(n): + if A >> j & 1: + marg_num[j] += v + in_regime = all(mn * RECORD_DEN < RECORD_NUM * Z for mn in marg_num) + jmax = max(range(n), key=lambda j: marg_num[j]) + max_marg_float = marg_num[jmax] / Z + # entropy of mu (float, reporting only) + H = -sum((v / Z) * math.log2(v / Z) for v in mu_row if v) + + # census + num_terms: dict[Fraction, int] = {} # q = OR/t -> sum of masses + W_tot = 0 # sum_{i 0 and t01 > 0 and t10 > 0 and t11 > 0 + if nondeg != (a in active and b in active): + dich_ok = False + if nondeg: + m = t00 + t01 + t10 + t11 + q = Fraction(t00 * t11, t01 * t10) / t + rows_i.append((m, q)) + w_i += m + if i < n: + for m, q in rows_i: + num_terms[q] = num_terms.get(q, 0) + m + W_tot += w_i + n_rows += len(rows_i) + else: + w_last = w_i + # free anchor (005 Prop 1): OR = t exactly at every + # nondegenerate history at i = n, i.e. q = OR/t = 1 + prop1_ok = all(q == 1 for _m, q in rows_i) + per_i.append({"i": i, "n_active": len(active), + "n_rows": len(rows_i), "w_i_over_Z": w_i / Z if Z else 0}) + if progress and i % 16 == 0: + print(f" ... census i={i}/{n} " + f"({time.monotonic() - t_start:.0f}s)", flush=True) + + # certified numerator enclosure (deduplicated log2 calls) + N_lo = N_hi = F0 + for q, m in num_terms.items(): + lo, hi = log2_enclosure(q) + N_lo += m * lo + N_hi += m * hi + agg_lo = N_lo / W_tot + agg_hi = N_hi / W_tot + # alternative bookkeeping (013): every degenerate positive-mass history + # scored at exactly lambda -> per-i defined mass becomes the full + # history mass Z, numerator unchanged: A_alt = N / ((n-1) Z). + alt_lo = N_lo / ((n - 1) * Z) + alt_hi = N_hi / ((n - 1) * Z) + # i = n inclusion variant: M_n = lambda identically, so only the + # denominator grows. + incl_lo = N_lo / (W_tot + w_last) + incl_hi = N_hi / (W_tot + w_last) + # union-closedness of the support (informational: the control statement + # quantifies over ALL mu, closure is not a hypothesis -- checked so the + # record can say so explicitly) + sset = set(supp) + uc = all((A | B) in sset for A in supp for B in supp) + return { + "n": n, "k_atoms": m_atoms, "t": t, + "agg_lo": agg_lo, "agg_hi": agg_hi, + "alt_lo": alt_lo, "alt_hi": alt_hi, + "incl_lo": incl_lo, "incl_hi": incl_hi, + "width": agg_hi - agg_lo, + "W_over_Z": W_tot / Z, "dich_ok": dich_ok, "prop1_ok": prop1_ok, + "in_regime": in_regime, "max_marg_float": max_marg_float, + "argmax_marg_coord": jmax + 1, "H_bits": H, + "n_log_terms": len(num_terms), "n_rows": n_rows, + "union_closed_support": uc, + "per_i": per_i, + "secs": round(time.monotonic() - t_start, 1), + } + + +# ====================================================================== +# instance builders (re-implemented from 014's prose; data from +# checkpoints) +# ====================================================================== + +def mu_ladder_from_theta(n: int, r: int, th: dict[str, float] + ) -> dict[int, Fraction]: + """MU(n, r) from the prose of 014 section 1. Coordinate j = bit j-1. + Block a: {1,n-1}:A1, {1,n}:A2, light slices {1,c_j,n}:light; + block b: {}:B0, {n-1}:B1, {n}:B2, responders {c_j,n-1}:resp; + dilution {2},{3},{4}:dil; response coordinates c_j = 5..4+r. + Weights taken as EXACT dyadic rationals of the checkpoint floats.""" + assert 1 <= r <= n - 6 + hi = 1 << (n - 1) # coordinate n + lo = 1 << (n - 2) # coordinate n-1 + F = Fraction + u = { + 1 | lo: F(th["A1"]), 1 | hi: F(th["A2"]), 0: F(th["B0"]), + lo: F(th["B1"]), hi: F(th["B2"]), + } + for j in range(r): + c = 1 << (4 + j) # coordinate 5+j + u[1 | c | hi] = F(th["light"]) + u[c | lo] = F(th["resp"]) + for d in (1, 2, 3): # coordinates 2, 3, 4 + u[1 << d] = F(th["dil"]) + assert len(u) == 2 * r + 8 + return u + + +def tidy(u: dict[int, Fraction], den: int = 10 ** 4) -> dict[int, Fraction]: + """014's tidy-rational convention: limit_denominator(1e4), zeros + dropped after rationalization.""" + out = {a: w.limit_denominator(den) for a, w in u.items()} + return {a: w for a, w in out.items() if w > 0} + + +# 007's 10-atom witness (weights are DATA copied from 007's record / +# checkpoint; n = 7, response coordinate 5, futures 6 and 7) +WITNESS_THETA = { + "A1": 3.802705744653898, "A2": 8.094727225702503, + "light": 0.004792968160237178, "B0": 0.2526660836353666, + "B1": 27.28641199464732, "B2": 12.529780963546393, + "resp": 0.3609234435898425, "dil": 20.0, +} + + +def fmt(r: dict) -> str: + return (f"A in [{float(r['agg_lo']):+.12f}, {float(r['agg_hi']):+.12f}] " + f"(width {float(r['width']):.1e}); " + f"{'KILL' if r['agg_hi'] < 0 else 'POSITIVE' if r['agg_lo'] > 0 else 'STRADDLE'}; " + f"in-regime {r['in_regime']} (max marg {r['max_marg_float']:.5f} " + f"at coord {r['argmax_marg_coord']}); dich {r['dich_ok']}; " + f"prop1 {r['prop1_ok']}; " + f"H {r['H_bits']:.3f}; W/Z {float(r['W_over_Z']):.4f}; " + f"alt-conv [{float(r['alt_lo']):+.6f}, {float(r['alt_hi']):+.6f}]; " + f"union-closed {r['union_closed_support']}; {r['secs']}s") + + +def save(name: str, obj) -> None: + DATA.mkdir(exist_ok=True) + + def default(o): + if isinstance(o, Fraction): + return str(o) if o.denominator < 10 ** 40 else repr(float(o)) + return str(o) + (DATA / name).write_text(json.dumps(obj, indent=1, sort_keys=True, + default=default) + "\n", + encoding="utf-8") + + +def public(r: dict) -> dict: + keep = ("n", "k_atoms", "in_regime", "max_marg_float", + "argmax_marg_coord", "H_bits", "dich_ok", "prop1_ok", + "n_log_terms", + "n_rows", "union_closed_support", "secs") + out = {k: r[k] for k in keep} + out["t"] = str(r["t"]) + for k in ("agg_lo", "agg_hi", "alt_lo", "alt_hi", "incl_lo", "incl_hi", + "width", "W_over_Z"): + out[k + "_float"] = float(r[k]) + out["verdict"] = ("CERTIFIED NEGATIVE (KILL)" if r["agg_hi"] < 0 else + "CERTIFIED POSITIVE" if r["agg_lo"] > 0 else + "STRADDLES 0") + return out + + +def main() -> None: + quick = "--quick" in sys.argv + t0 = time.monotonic() + + print("[S0] log2 enclosure self-test (attack g)") + st = _selftest_log2() + for c in st["cases"]: + print(f" q={c['q'][:40]:40s} ref={c['ref']:+.12f} " + f"[{c['lo']:+.12f},{c['hi']:+.12f}] w={c['width']:.1e} " + f"{'ok' if c['ok'] else 'FAIL'}") + print(f" additivity {st['additivity_ok']}, pow2 exact " + f"{st['pow2_exact']}, ALL {'OK' if st['ok'] else 'FAIL'}") + save("aggsk15_selftest.json", st) + assert st["ok"], "log2 enclosure self-test failed" + + results = {} + + # ---- anchor: 007's witness at t = 181/16 (013 part C: +1.844669005) + uw = mu_ladder_from_theta(7, 1, WITNESS_THETA) + r = exact_aggregate(uw, 7, Fraction(181, 16)) + print("[S1] witness n=7 t=181/16 (expect +1.844669005):") + print(" " + fmt(r)) + results["witness_t181_16"] = public(r) + + # ---- the load-bearing kill: theta*(2.0) ladder at n = 96, t = 4 + d = json.loads((DATA / "aggprobe2_partP2.json").read_text()) + th96 = [row for row in d["rows"] if row["stage"] == "transfer" + and row["lambda"] == 2.0 and row["n"] == 96][0]["theta"] + u96 = mu_ladder_from_theta(96, 90, th96) + print(f"[S2] theta* ladder n=96 r=90 ({len(u96)} atoms) t=4 " + f"(014 cert: [-0.000759865185844166087, ...086]):") + r96 = exact_aggregate(u96, 96, Fraction(4), progress=True) + print(" " + fmt(r96)) + results["theta2_n96_t4"] = public(r96) + save("aggsk15_results.json", results) + + # ---- positive control at n = 32: P3-best, tidy zero-dropped, t = 4 + if not quick: + d3 = json.loads((DATA / "aggprobe2_partP3.json").read_text()) + b = d3["best"] + u32 = tidy({int(k, 2): Fraction(v) for k, v in b["u"].items()}) + print(f"[S3] P3-best n={b['n']} tidy ({len(u32)} atoms) t=4 " + f"(014 cert: +0.297679573):") + r32 = exact_aggregate(u32, b["n"], Fraction(4)) + print(" " + fmt(r32)) + results["p3best_n32_tidy_t4"] = public(r32) + save("aggsk15_results.json", results) + + # ---- surviving-positive-space check: RAW witness-weight ladder at + # n = 96 with 014's P6 adversarial dilution weight, t = 4 + if not quick: + d6 = json.loads((DATA / "aggprobe2_partP6.json").read_text()) + row6 = [x for x in d6["rows"] if x["n"] == 96 + and x["lambda"] == 2.0][0] + thraw = dict(WITNESS_THETA) + thraw["dil"] = row6["wdil"] + uraw = mu_ladder_from_theta(96, 90, thraw) + print(f"[S4] RAW witness ladder n=96 wdil={row6['wdil']:.6f} t=4 " + f"(014 P6 float: {row6['agg']:+.6f}):") + rr = exact_aggregate(uraw, 96, Fraction(4), progress=True) + print(" " + fmt(rr)) + results["raw_ladder_n96_t4"] = public(rr) + save("aggsk15_results.json", results) + + # ---- optional deeper kills + if "--n128" in sys.argv: + th128 = [row for row in d["rows"] if row["stage"] == "transfer" + and row["lambda"] == 2.0 and row["n"] == 128][0]["theta"] + u128 = tidy(mu_ladder_from_theta(128, 122, th128)) + print(f"[S5] theta* ladder n=128 tidy ({len(u128)} atoms) t=4 " + f"(014 cert: -0.015405727):") + r128 = exact_aggregate(u128, 128, Fraction(4), progress=True) + print(" " + fmt(r128)) + results["theta2_n128_tidy_t4"] = public(r128) + save("aggsk15_results.json", results) + if "--n160" in sys.argv: + d7 = json.loads((DATA / "aggprobe2_partP7.json").read_text()) + u160 = {int(k, 2): Fraction(float(v)) + for k, v in d7["u_target"].items()} + print(f"[S6] theta* ladder n=160 dyadic ({len(u160)} atoms) t=4 " + f"(014 cert: -0.024225909):") + r160 = exact_aggregate(u160, 160, Fraction(4), progress=True) + print(" " + fmt(r160)) + results["theta2_n160_dyadic_t4"] = public(r160) + save("aggsk15_results.json", results) + + print(f"done in {time.monotonic() - t0:.0f}s") + + +if __name__ == "__main__": + main() diff --git a/problems/union-closed/prior-art.json b/problems/union-closed/prior-art.json index b85eda7..79767ad 100644 --- a/problems/union-closed/prior-art.json +++ b/problems/union-closed/prior-art.json @@ -1,480 +1,519 @@ { - "problem": "union-closed", - "route_status": "LIVE", - "route_summary": "Dependent-coupling interface survives every known counterexample genre; model ceiling 0.4315 vs published record 0.38271. No bound claimed; three labeled proof gaps.", - "attempts": [ - { - "id": "001", - "file": "attempts/001-entropy-barrier-map.md", - "date": "2026-07-25", - "mode": "informed", - "status": "MAP", - "mechanism": [ - "entropy-method", - "barrier-analysis", - "exhaustive-small-family-search", - "k-wise-union", - "up-sets" - ], - "one_line": "Mapped the entropy-method frontier and the (3-sqrt5)/2 barrier; established the n<=4 exhaustive baseline; produced candidate ideas A-D.", - "refuted_by": null, - "gaps": [], - "leak_terms": [ - "k-wise union", - "idea A", - "idea B", - "idea C", - "idea D", - "p(c) ladder" - ] - }, - { - "id": "002", - "file": "attempts/002-weighted-kl-ladder.md", - "date": "2026-07-25", - "mode": "informed", - "status": "REFUTED", - "mechanism": [ - "weighted-kl", - "entropy-functional-reweighting", - "smoothing-insensitive-functional" - ], - "one_line": "The weighted-KL ladder is dead for every c >= 0, killed by Sawin geometric mixtures; generalizes to a no-go covering all smoothing-insensitive functionals.", - "refuted_by": "sawin-geometric-mixtures", - "gaps": [], - "leak_terms": [ - "weighted-KL", - "weighted KL ladder", - "smoothing-insensitive" - ] - }, - { - "id": "003", - "file": "attempts/003-dependent-couplings.md", - "date": "2026-07-26", - "mode": "informed", - "status": "LIVE", - "superseded_by": "004", - "mechanism": [ - "coupling", - "overlap-tilt", - "sinkhorn", - "plackett", - "conditionally-iid", - "block-adaptive-coupling" - ], - "one_line": "Overlap-biased couplings separate every adversary that killed the KL ladder; evades the 002 no-go by being smoothing-sensitive. No bound claimed.", - "refuted_by": null, - "gaps": [ - "plackett-odds-ratio-control", - "mutual-information-tax", - "recipe-totality" - ], - "leak_terms": [ - "overlap-tilt", - "overlap-biased", - "Sinkhorn", - "Plackett", - "dependent coupling", - "conditionally-iid" - ] - }, - { - "id": "004", - "file": "attempts/004-skeptic-review-of-003.md", - "date": "2026-07-26", - "mode": "informed", - "status": "VERIFIED_WITH_CORRECTIONS", - "verifies": "003", - "mechanism": [ - "adversarial-verification", - "independent-reimplementation", - "coupling", - "overlap-tilt" - ], - "one_line": "Independent re-implementation confirms 003's interface and separations are real; corrects the recipe ceiling to 0.431496 and refutes 003's mini-theorem (e) as stated.", - "refuted_by": null, - "gaps": [ - "plackett-odds-ratio-control", - "mutual-information-tax", - "recipe-totality" - ], - "corrections": [ - "mini-theorem (e) bracket positivity is false as stated; restated, with the {0} u [1/2,1) genre covered by a half-mixing coupling", - "single-parameter recipe ceiling is 0.431496 in closed form, not 0.445 (grid-floor artifact)", - "one non-reproducible data file; several overstated sentences" - ], - "leak_terms": [ - "0.4315", - "0.431496", - "half-mixing", - "mini-theorem" - ] - }, - { - "id": "005", - "file": "attempts/005-odds-ratio-control-refuted.md", - "date": "2026-07-30", - "mode": "informed", - "status": "REFUTED", - "mechanism": [ - "coupling", - "overlap-tilt", - "plackett", - "sinkhorn", - "total-positivity", - "barrier-analysis" - ], - "one_line": "The Gap-1 pointwise odds-ratio control [1, 2^lambda] is false both ways: Cauchy-Schwarz forces OR >= 2^lambda on diagonal histories, and an explicit 4-atom family crashes OR to 2^{lambda(3-n)}; sharp range and MTP2 one-sided partial result proved. Route stays LIVE with Gap 1 restated.", - "leak_terms": [ - "odds-ratio crash", - "crash family", - "future-overlap discrepancy", - "over-diagonalizes", - "2^{lambda(3-n)}", - "averaged-odds-ratio-control", - "OR census", - "uc_odds_ratio" - ], - "refuted_by": null, - "gaps": [ - "averaged-odds-ratio-control", - "mutual-information-tax", - "recipe-totality" - ] - }, - { - "id": "006", - "file": "attempts/006-skeptic-review-of-005.md", - "date": "2026-07-30", - "mode": "informed", - "status": "VERIFIED_WITH_CORRECTIONS", - "verifies": "005", - "mechanism": [ - "adversarial-verification", - "independent-reimplementation", - "coupling", - "overlap-tilt", - "sinkhorn", - "total-positivity" - ], - "one_line": "Independent re-implementation and hand re-derivation confirm all six propositions of 005 including the crash identity at lambdas 005 never ran; the Karlin-Rinott step is checked against the literature and re-proved via Ahlswede-Daykin; corrections are minor (Prop 2 equality-iff needs lambda > 0, two part-B reporting slips, docstring bugs); the restated averaged-control gap survives its own first falsifiable test.", - "corrections": [ - "Prop 2 equality characterization ('equality iff proportional slices') is false at lambda = 0, where OR = 1 identically for every mu; the iff needs lambda > 0", - "part-B prose misreports its checkpoint: the 1.0000000011 minimum is the i= 1' where code and record prove OR >= 2^lambda; part G says overall deviation scales like delta where the data show delta^2", - "'about 83% of history mass' is the lambda = 0.5 figure; it is 81% at lambda = 1.5" - ], - "refuted_by": null, - "gaps": [ - "averaged-odds-ratio-control", - "mutual-information-tax", - "recipe-totality" - ], - "leak_terms": [ - "M_i census", - "averaged control survives crash", - "family-dependent delta-squared onset", - "uc_or_skeptic" - ] - }, - { - "id": "007", - "file": "attempts/007-averaged-or-control.md", - "date": "2026-07-31", - "mode": "informed", - "status": "REFUTED", - "mechanism": [ - "coupling", - "overlap-tilt", - "plackett", - "sinkhorn", - "total-positivity", - "gram-mass-pairing" - ], - "one_line": "The restated Gap-1 averaged control M_i >= lambda is made well-posed (forced degeneracy bookkeeping), proved for four structured subclasses, then refuted in the record-relevant regime by an explicit perturbation-stable 10-atom mu (marginals <= 0.318) with M_5 = lambda - 0.122; the i-aggregated form survives and becomes the new gap.", - "range": "444-instance census plus clamped adversarial searches at n <= 8, lambda in [0.05, 6]; witness verified by two independent engines to 1e-14, stable under 3% perturbation and 2-digit rounding, violating for every lambda >= 0.03 on the same mu", - "refuted_by": null, - "gaps": [ - "aggregated-or-control", - "margin-modulated-or-control", - "mutual-information-tax", - "recipe-totality" - ], - "leak_terms": [ - "two-prefix witness", - "light slice", - "light-slice", - "10-atom witness", - "0.122033", - "dilution invariance", - "Gram mass pairing", - "gram-mass-pairing", - "2x2-minor control", - "cell floor", - "degeneracy dichotomy", - "aggregated-or-control", - "uc_or_avg", - "or_avg_part" - ], - "methodological_lesson": "Cross-engine agreement is not independence when both engines share IEEE arithmetic: a 1e-299 atom made u(A)u(B) underflow to exact zero in one diagonal cell, and both implementations plus the Sinkhorn round-trip agreed on the resulting fake violation; a parameter sweep of the suspect light atom toward its boundary was what exposed it. Guard adopted: cell-floor diagnostic plus dynamic-range clamping in searches.", - "note": "Settles 005 lead 1 negatively (005's SPECULATION label was accurate; nothing in 005/006 needs correction). Route stays LIVE: this kills a candidate bridge, not the interface. Partial positives now proved: M_i >= lambda for supports of <= 4 atoms, for potential-MTP2 mu, at i in {1, n}, and to first order in lambda (perfect-square coefficient), the last modulo the standard Sinkhorn-smoothness step." - }, - { - "id": "008", - "file": "attempts/008-perturbative-assembly.md", - "date": "2026-07-31", - "mode": "informed", - "status": "LIVE", - "mechanism": [ - "coupling", - "overlap-tilt", - "sinkhorn", - "plackett", - "perturbative-expansion", - "implicit-function-theorem" - ], - "one_line": "Closes the missing smoothness step in 005 Prop 6 (quantitative IFT, explicit fixed-n budget 924*min(p,1-p)^{-3n}*delta^2), runs the perturbative assembly end-to-end (positive on every marginal-capped instance to delta=0.2, n<=9), and shows the pointwise delta^2 OR budget is NOT n-uniform (deviation grows ~0.3 bits/coordinate at fixed delta) while the coordinate-averaged downward budget flattens; conditional n-uniform assembly theorem with every budget labelled by its gap.", - "range": "Exact couplings at n <= 9; delta <= 0.2; directions crash / zero-prefix embedded crash / delta_empty / Bernoulli mixtures / slice / pm-mixture; (p, rho) = (0.383, 1.06|1.20) and (0.30, 2.0); 78 assembly runs, 80 budget-census fits; proofs (Lemmas A-C, Theorem P6', conditional theorem) are for all n but pending skeptic pass", - "refuted_by": null, - "gaps": [ - "averaged-odds-ratio-control", - "mutual-information-tax", - "recipe-totality", - "sinkhorn-tv-stability" - ], - "leak_terms": [ - "perturbative assembly", - "capped delta-ball", - "downward budget", - "centered-overlap", - "zero-prefix embedded crash", - "n-uniform budget", - "uc_pert", - "Theorem P6", - "tau_half", - "budget census" - ], - "note": "Sub-results at different confidence levels inside one record: the Prop-6 closure is a written-out proof (VERIFIED pending independent skeptic pass); the lead-3 assembly check and the non-uniformity of the pointwise budget are EVIDENCE at n <= 9; the assembly theorem is CONDITIONAL on budgets (B1)-(B4) with n-uniform constants, of which (B1) is the perturbative form of gap a-prime and (B2) of gap b. 005's leads 3 and 4 are distinct (the task line conflated them); both were run." - }, - { - "id": "009", - "file": "attempts/009-mutual-information-tax.md", - "date": "2026-07-31", - "mode": "informed", - "status": "EVIDENCE", - "mechanism": [ - "coupling", - "overlap-tilt", - "sinkhorn", - "chain-rule-assembly", - "mutual-information-tax" - ], - "one_line": "Gap 2 made precise as a chain-rule assembly statement (TAX at p) and probed: it survives all four adversary genres, is smoothing-sensitive on 002's own certificates, the pure slice tilt is provably tax-free, the half-mixing coupling provably loses nothing to chain-rule certification, and a previously unnamed second tax (measured ~0.40 log2 n on slices) is identified as the binding loss channel.", - "range": "Crash family n<=11, lambda in [-1,3]; half-mixing genre n<=14 at two parameter points; Sawin certificate gadgets n<=60000 (block recipe, rigorous bound); CL pure slice n<=80, lambda in [0,3], p in {0.42, 0.3823}. Finite-n, finite-grid EVIDENCE only; mini-lemmas SL/HM proved but not yet skeptic-reviewed.", - "refuted_by": null, - "gaps": [ - "averaged-odds-ratio-control", - "mutual-information-tax", - "recipe-totality", - "second-tax-scaling" - ], - "leak_terms": [ - "chain-rule assembly value", - "second tax", - "(TAX at p)", - "tax-free slice", - "uc_mitax", - "0.40 log2 n", - "lambda-window narrowing", - "CR = Gain identically" - ], - "note": "Deliberately the smallest-budget line of the cycle: a probe, not a campaign. The expected outcome was a precise obstruction; instead the candidate survived everywhere tested and the losses were localized (T_A vs ST, by genre)." - }, - { - "id": "010", - "file": "attempts/010-ideation-sweep-fields.md", - "date": "2026-07-31", - "mode": "informed", - "status": "MAP", - "mechanism": [ - "field-lens-sweep", - "barrier-analysis" - ], - "one_line": "First ideation sweep (6 field lenses): 3 queue-worthy routes (pairwise-closure LP certification, union-transfer-operator eigenvalue field, bipartite MIS decomposition), 6 secondary, 5 no-purchase verdicts incl. a proven product-weight no-go killing the multiplicative toolbox.", - "leak_terms": [ - "field-lens sweep", - "coatom eigenvalue", - "union transfer operator", - "eigenvalue field", - "log-supermodular lambda", - "product-weight no-go", - "pairwise-closure LP", - "union-injection Hall", - "union-escape graph", - "MIS share r(G)", - "crosscut truncation" - ], - "refuted_by": null, - "gaps": [] - }, - { - "id": "011", - "file": "attempts/011-skeptic-review-of-009.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED_WITH_CORRECTIONS", - "verifies": "009", - "mechanism": [ - "adversarial-verification", - "independent-reimplementation", - "coupling", - "overlap-tilt", - "sinkhorn", - "chain-rule-assembly", - "mutual-information-tax" - ], - "one_line": "Independent re-implementation (recursion engine, natural-log Sinkhorn, direct-MI cross-checks) confirms 009 in full: mini-lemmas SL and HM re-derived by hand and now VERIFIED, all headline numbers reproduce to <= 1.4e-10, the crash lambda-window sharpens to the law lam_max*(n-3) -> 4.8475, the ST ~ 0.40 log2 n fit survives extension to n = 240 and two new p, and the untested tilt-on-Sawin cell closes positive at n <= 300; corrections are reporting-level (grid-edge suprema quoted as suprema, mixed-p 'increasing margin', slice-ratio scope, a measured 1e-5 lgamma floor behind part D's 'exactly', and an unstated m_k = 1/2 legality hypothesis p_k >= 1-1/sqrt(2) in the block bound, verified to hold on every instance used).", - "corrections": [ - "slice sup_lambda CR values at p ~ 0.42 are grid-edge values; true suprema sit at lambda ~ 4-5 and are larger (e.g. +1.4655 vs reported +1.0199 at n=80, w0=34); 'increasing margin in n' conflates p = 0.4167 and p = 0.425 rows -- at fixed p the margin is cleanly increasing", - "Outcome scope 'p in {0.42, 0.3823}' overstates: slice ratios actually run are {0.4167, 0.425, 0.3833, 0.3875}; p = 0.3823 itself was never tested", - "part D's 'H(mu) computed exactly' has a measured ~4e-6 absolute lgamma floor at n = 60000 (harmless at the +2418 scale)", - "the part-D block bound's m_k = 1/2 choice needs the unstated hypothesis p_k >= 1-1/sqrt(2) ~ 0.29289; holds on all instances used but would fail silently on mixtures with a component marginal below it" - ], - "refuted_by": null, - "gaps": [ - "averaged-odds-ratio-control", - "mutual-information-tax", - "recipe-totality", - "second-tax-scaling" - ], - "leak_terms": [ - "uc_mitax_skeptic", - "mitaxsk", - "lambda-window law", - "4.847/(n-3)", - "0.405 log2 n", - "tilt-CR on Sawin", - "recursion engine", - "most conservative assembly", - "m_k legality" - ] - }, - { - "id": "012", - "file": "attempts/012-skeptic-review-of-008.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED_WITH_CORRECTIONS", - "verifies": "008", - "mechanism": [ - "adversarial-verification", - "independent-reimplementation", - "coupling", - "overlap-tilt", - "sinkhorn", - "perturbative-expansion", - "implicit-function-theorem" - ], - "one_line": "Line-by-line re-derivation and two independent engines (incl. an orbit-symmetrized census that reaches n = 32) confirm 008's Theorem P6' closure of the Prop-6 smoothness gap, the 78-run assembly table, the calibration block, and the centered-kernel closed form; but the conditional theorem's constants fail (box-uniform tau_half is analytically false at 008's own s0 = 0.10, ceiling s0 <= 0.0843; delta_0 = 0.022 should be <= 0.0079 by 008's own formula, ~0.004 corrected), and the extended census reverses the flattening reading: the averaged downward OR budget's increments stop shrinking and start growing past n ~ 22 (claimed limits 0.07/0.3 passed by n = 16/10, delta^2-coefficient verified stable in delta at n = 32), killing budget (B1) as stated at evidence level by 008's own lead-1 criterion, while the tax budget (B2) is first-order in delta, not quadratic, in every observable window.", - "corrections": [ - "conditional theorem's 'G nondecreasing in tau up to tau_half on the box' is analytically false at (p, rho, s0) = (0.383, 1.2, 0.10): (x0+s0)^2 = 0.5141 > 1/2 and z > xy for rho > 1, so tau_half(corner) = 0; the claimed P2d numerical check does not exist in uc_pert.py; repair requires s0 <= x_c - x0 = 0.0843 with re-measured constants", - "delta_0 ~ 0.022 does not follow from 008's stated formula: the code drops the ||G||_box/s0^2 c3 term (||G||_box = 0.1456, never computed by 008 -> delta_0 = 0.0079); the off-box charge needs the global sup(-G) = 0.487, not ||G||_box; M_tau must include the tau < 0 side (0.0838 vs 0.0592); c3 is ~4.7 not 2.5; fully corrected delta_0 ~ 0.004", - "the averaged downward budget does NOT flatten: at (0.383, 1.2) the delta->0 coefficient rises 0.065 (n=9) -> 0.104 (n=32) with increasing increments; at (0.30, 1.0) it passes the claimed 0.3 limit at n = 10 and reaches >= 1.24 by n = 32 (the fixed-delta turnover is a delta-saturation artifact); (B1) with n-uniform constant is dead at EVIDENCE level per 008's own lead-1 falsification criterion", - "the tax is first-order in delta (halving ratios -> 2), not O(delta^2), in every observable window at large n; fixed-delta tax/delta^2 exceeds the instantiated c2 = 0.01 by 10x at n = 32; (B2) must be restated as a linear-in-delta budget (absolute tax stays tiny, so end-to-end positivity is not threatened)", - "'worst in-class net = +0.0003 (slice n=5 delta=0.2)' is out of class by 008's own capped flag; true worst in-class min-net is ~ +0.00105 (rho=1.06 crash n=6 delta=0.02)", - "P6' constant chase: the stated step 'using m0 <= 1/2' yields 831, not 924 (924 corresponds to m0 <= 1; bound valid either way); the 'measured c ~ 0.6' looseness illustration ignores 008's own census worst case c ~ 20.6" - ], - "range": "Hand re-derivation of Lemmas A-C and Theorem P6' in full; P6' bound stress-tested on 120 instances (n <= 6, five directions incl. delta_full and half-empty, delta up to the theorem boundary 0.5, zero violations); Lemma B checked at interior delta; 78-run table spot-verified to 5e-6 and 104-run capped-negative hunt (rho in {1.045, 1.06, 1.20}, n <= 9, delta <= 0.2, incl. the half-empty direction with marginals exactly p) found none; budget census re-run and extended to n = 32 on the crash direction via an orbit-symmetrized engine (n <= 9 values reproduce 008 to all printed digits), with delta -> 0 stability verified at n = 32 down to delta = 0.003125 for the downward budget (clean delta^2) and shown NOT to hold for the tax (delta-linear); conditional-theorem constants recomputed on 008's grid plus tau < 0 and global domains", - "gaps": [ - "averaged-odds-ratio-control", - "mutual-information-tax", - "recipe-totality", - "sinkhorn-tv-stability" - ], - "leak_terms": [ - "uc_pert_skeptic", - "pertsk", - "orbit engine", - "budget growth resumes", - "delta-linear tax", - "s0 ceiling 0.0843", - "tau_half corner", - "corrected delta0", - "n = 32 census" - ] - }, - { - "id": "013", - "file": "attempts/013-skeptic-review-of-007.md", - "date": "2026-07-31", - "mode": "informed", - "status": "VERIFIED", - "verifies": "007", - "mechanism": [ - "adversarial-verification", - "independent-reimplementation", - "exact-rational-arithmetic", - "coupling", - "overlap-tilt", - "sinkhorn", - "gram-mass-pairing" - ], - "one_line": "Exact rational arithmetic (Fraction end-to-end, certified log2 enclosures) confirms 007 in full: the 10-atom witness's violation is certified float-free at rational tilts spanning lambda ~ 0.0144..5 including exactly rational lambda = 2, 3, 4 (positive at small t exactly where Theorem C forces it), both invariances hold as exact rational identities, the aggregate is certified positive (+1.844669) and survives witness-seeded two-coordinate attacks, all four partial theorems re-derive by hand, and the refutation even survives the conjecture-friendliest alternative bookkeeping (degenerate histories scored at lambda: still -0.067); no corrections, one cosmetic runtime discrepancy.", - "range": "Witness certified at 12 exact rational tilt parameters (fixed-u family t = 101/100 .. 32 plus rationalized-refit measures at lambda = 1, 2, 3 within 9.2e-16 of the witness mu); enclosure widths < 1e-40; own 96-instance battery + 21-endpoint aggregate hunt at n <= 7; Theorem B probed on 500 random <= 4-atom instances; hand re-derivations of section 1, Lemmas G/M, Theorems A/B/C and both invariances", - "refuted_by": null, - "gaps": [ - "aggregated-or-control", - "margin-modulated-or-control", - "mutual-information-tax", - "recipe-totality" - ], - "leak_terms": [ - "oravgsk", - "uc_or_avg_skeptic", - "exact rational witness", - "convention robustness", - "surrogate-lambda", - "1.844669", - "minor ratio 6.68" - ], - "note": "Decisive check: 007's confessed vulnerability (both engines share IEEE floats; its first witnesses were shared-underflow artifacts) is closed by re-verifying the witness in exact rational arithmetic with directed-rounding log2 enclosures. At t = 4 and t = 16 the refutation of M_i >= lambda is a fully rational statement with lambda = 2, 4 exactly. New strengthening beyond 007: the violation does not hinge on the conditioning-out convention (degenerate histories scored at exactly lambda still give M_5 - lambda = -0.067, certified). Nuance: exact certificates at lambda = 1, 2, 3 are for rationalized-refit measures within 9.2e-16 of the witness mu, not the identical rational mu. Theorem C's smoothness step is analogous to but not literally discharged by 008's P6-prime (P6-prime is d/d-mu at fixed lambda; Theorem C needs d/d-lambda at fixed mu)." - }, - { - "id": "014", - "file": "attempts/014-aggregated-control-probe-large-n.md", - "date": "2026-08-04", - "mode": "informed", - "status": "REFUTED", - "mechanism": [ - "coupling", - "overlap-tilt", - "sinkhorn", - "plackett", - "gram-mass-pairing", - "exact-rational-arithmetic" - ], - "one_line": "Large-n probe of the restated Gap 1 kills it: unit-replicating the 10-atom witness genre (r crash units sharing one block marker) with re-optimized shared weights drives the i-aggregated control negative past n ~ 90, certified float-free at t = 4 (n = 96 and 160, marginals <= 0.309); the n <= 32 space stays positive everywhere searched.", - "leak_terms": [ - "unit replication", - "multi-unit witness ladder", - "MU(n,r)", - "theta-star ladder", - "aggregate crossing", - "O(1) surplus corridor", - "0.024226", - "-0.000759865", - "uc_or_agg_probe", - "aggprobe" - ], - "range": "Free-support seeded climbs n <= 32 (58 climbs round 1, deep annealed round 2) plus permutation orbits: 0 violations; unit-symmetric ladder evaluated to n = 160 (raw witness weights: positive through n = 128, min +0.157; optimized theta*: float crossing in (88, 96], -0.0242 at n = 160, 20/20 perturbations stay negative); exact rational certificates at t = 181/16 and t = 4, enclosure widths < 1e-18", - "refuted_by": null, - "gaps": [ - "margin-modulated-or-control", - "mutual-information-tax", - "recipe-totality" - ], - "note": "Settles queue item 1's probe-first branch negatively: the aggregated control is certified positive at n <= 32 on everything searched, so proof effort there was exactly the small-n trap 012 warned about. The kill is convention-robust (scoring degenerate histories at lambda only grows the denominator, never the numerator). Route stays LIVE; Gap 1 falls to the margin-modulated candidate (007 lead 2), which this record's ladder family must now be run against." - } - ] -} \ No newline at end of file + "problem": "union-closed", + "route_status": "LIVE", + "route_summary": "Dependent-coupling interface survives every known counterexample genre; model ceiling 0.4315 vs published record 0.38271. No bound claimed; three labeled proof gaps.", + "attempts": [ + { + "id": "001", + "file": "attempts/001-entropy-barrier-map.md", + "date": "2026-07-25", + "mode": "informed", + "status": "MAP", + "mechanism": [ + "entropy-method", + "barrier-analysis", + "exhaustive-small-family-search", + "k-wise-union", + "up-sets" + ], + "one_line": "Mapped the entropy-method frontier and the (3-sqrt5)/2 barrier; established the n<=4 exhaustive baseline; produced candidate ideas A-D.", + "refuted_by": null, + "gaps": [], + "leak_terms": [ + "k-wise union", + "idea A", + "idea B", + "idea C", + "idea D", + "p(c) ladder" + ] + }, + { + "id": "002", + "file": "attempts/002-weighted-kl-ladder.md", + "date": "2026-07-25", + "mode": "informed", + "status": "REFUTED", + "mechanism": [ + "weighted-kl", + "entropy-functional-reweighting", + "smoothing-insensitive-functional" + ], + "one_line": "The weighted-KL ladder is dead for every c >= 0, killed by Sawin geometric mixtures; generalizes to a no-go covering all smoothing-insensitive functionals.", + "refuted_by": "sawin-geometric-mixtures", + "gaps": [], + "leak_terms": [ + "weighted-KL", + "weighted KL ladder", + "smoothing-insensitive" + ] + }, + { + "id": "003", + "file": "attempts/003-dependent-couplings.md", + "date": "2026-07-26", + "mode": "informed", + "status": "LIVE", + "superseded_by": "004", + "mechanism": [ + "coupling", + "overlap-tilt", + "sinkhorn", + "plackett", + "conditionally-iid", + "block-adaptive-coupling" + ], + "one_line": "Overlap-biased couplings separate every adversary that killed the KL ladder; evades the 002 no-go by being smoothing-sensitive. No bound claimed.", + "refuted_by": null, + "gaps": [ + "plackett-odds-ratio-control", + "mutual-information-tax", + "recipe-totality" + ], + "leak_terms": [ + "overlap-tilt", + "overlap-biased", + "Sinkhorn", + "Plackett", + "dependent coupling", + "conditionally-iid" + ] + }, + { + "id": "004", + "file": "attempts/004-skeptic-review-of-003.md", + "date": "2026-07-26", + "mode": "informed", + "status": "VERIFIED_WITH_CORRECTIONS", + "verifies": "003", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "coupling", + "overlap-tilt" + ], + "one_line": "Independent re-implementation confirms 003's interface and separations are real; corrects the recipe ceiling to 0.431496 and refutes 003's mini-theorem (e) as stated.", + "refuted_by": null, + "gaps": [ + "plackett-odds-ratio-control", + "mutual-information-tax", + "recipe-totality" + ], + "corrections": [ + "mini-theorem (e) bracket positivity is false as stated; restated, with the {0} u [1/2,1) genre covered by a half-mixing coupling", + "single-parameter recipe ceiling is 0.431496 in closed form, not 0.445 (grid-floor artifact)", + "one non-reproducible data file; several overstated sentences" + ], + "leak_terms": [ + "0.4315", + "0.431496", + "half-mixing", + "mini-theorem" + ] + }, + { + "id": "005", + "file": "attempts/005-odds-ratio-control-refuted.md", + "date": "2026-07-30", + "mode": "informed", + "status": "REFUTED", + "mechanism": [ + "coupling", + "overlap-tilt", + "plackett", + "sinkhorn", + "total-positivity", + "barrier-analysis" + ], + "one_line": "The Gap-1 pointwise odds-ratio control [1, 2^lambda] is false both ways: Cauchy-Schwarz forces OR >= 2^lambda on diagonal histories, and an explicit 4-atom family crashes OR to 2^{lambda(3-n)}; sharp range and MTP2 one-sided partial result proved. Route stays LIVE with Gap 1 restated.", + "leak_terms": [ + "odds-ratio crash", + "crash family", + "future-overlap discrepancy", + "over-diagonalizes", + "2^{lambda(3-n)}", + "averaged-odds-ratio-control", + "OR census", + "uc_odds_ratio" + ], + "refuted_by": null, + "gaps": [ + "averaged-odds-ratio-control", + "mutual-information-tax", + "recipe-totality" + ] + }, + { + "id": "006", + "file": "attempts/006-skeptic-review-of-005.md", + "date": "2026-07-30", + "mode": "informed", + "status": "VERIFIED_WITH_CORRECTIONS", + "verifies": "005", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "coupling", + "overlap-tilt", + "sinkhorn", + "total-positivity" + ], + "one_line": "Independent re-implementation and hand re-derivation confirm all six propositions of 005 including the crash identity at lambdas 005 never ran; the Karlin-Rinott step is checked against the literature and re-proved via Ahlswede-Daykin; corrections are minor (Prop 2 equality-iff needs lambda > 0, two part-B reporting slips, docstring bugs); the restated averaged-control gap survives its own first falsifiable test.", + "corrections": [ + "Prop 2 equality characterization ('equality iff proportional slices') is false at lambda = 0, where OR = 1 identically for every mu; the iff needs lambda > 0", + "part-B prose misreports its checkpoint: the 1.0000000011 minimum is the i= 1' where code and record prove OR >= 2^lambda; part G says overall deviation scales like delta where the data show delta^2", + "'about 83% of history mass' is the lambda = 0.5 figure; it is 81% at lambda = 1.5" + ], + "refuted_by": null, + "gaps": [ + "averaged-odds-ratio-control", + "mutual-information-tax", + "recipe-totality" + ], + "leak_terms": [ + "M_i census", + "averaged control survives crash", + "family-dependent delta-squared onset", + "uc_or_skeptic" + ] + }, + { + "id": "007", + "file": "attempts/007-averaged-or-control.md", + "date": "2026-07-31", + "mode": "informed", + "status": "REFUTED", + "mechanism": [ + "coupling", + "overlap-tilt", + "plackett", + "sinkhorn", + "total-positivity", + "gram-mass-pairing" + ], + "one_line": "The restated Gap-1 averaged control M_i >= lambda is made well-posed (forced degeneracy bookkeeping), proved for four structured subclasses, then refuted in the record-relevant regime by an explicit perturbation-stable 10-atom mu (marginals <= 0.318) with M_5 = lambda - 0.122; the i-aggregated form survives and becomes the new gap.", + "range": "444-instance census plus clamped adversarial searches at n <= 8, lambda in [0.05, 6]; witness verified by two independent engines to 1e-14, stable under 3% perturbation and 2-digit rounding, violating for every lambda >= 0.03 on the same mu", + "refuted_by": null, + "gaps": [ + "aggregated-or-control", + "margin-modulated-or-control", + "mutual-information-tax", + "recipe-totality" + ], + "leak_terms": [ + "two-prefix witness", + "light slice", + "light-slice", + "10-atom witness", + "0.122033", + "dilution invariance", + "Gram mass pairing", + "gram-mass-pairing", + "2x2-minor control", + "cell floor", + "degeneracy dichotomy", + "aggregated-or-control", + "uc_or_avg", + "or_avg_part" + ], + "methodological_lesson": "Cross-engine agreement is not independence when both engines share IEEE arithmetic: a 1e-299 atom made u(A)u(B) underflow to exact zero in one diagonal cell, and both implementations plus the Sinkhorn round-trip agreed on the resulting fake violation; a parameter sweep of the suspect light atom toward its boundary was what exposed it. Guard adopted: cell-floor diagnostic plus dynamic-range clamping in searches.", + "note": "Settles 005 lead 1 negatively (005's SPECULATION label was accurate; nothing in 005/006 needs correction). Route stays LIVE: this kills a candidate bridge, not the interface. Partial positives now proved: M_i >= lambda for supports of <= 4 atoms, for potential-MTP2 mu, at i in {1, n}, and to first order in lambda (perfect-square coefficient), the last modulo the standard Sinkhorn-smoothness step." + }, + { + "id": "008", + "file": "attempts/008-perturbative-assembly.md", + "date": "2026-07-31", + "mode": "informed", + "status": "LIVE", + "mechanism": [ + "coupling", + "overlap-tilt", + "sinkhorn", + "plackett", + "perturbative-expansion", + "implicit-function-theorem" + ], + "one_line": "Closes the missing smoothness step in 005 Prop 6 (quantitative IFT, explicit fixed-n budget 924*min(p,1-p)^{-3n}*delta^2), runs the perturbative assembly end-to-end (positive on every marginal-capped instance to delta=0.2, n<=9), and shows the pointwise delta^2 OR budget is NOT n-uniform (deviation grows ~0.3 bits/coordinate at fixed delta) while the coordinate-averaged downward budget flattens; conditional n-uniform assembly theorem with every budget labelled by its gap.", + "range": "Exact couplings at n <= 9; delta <= 0.2; directions crash / zero-prefix embedded crash / delta_empty / Bernoulli mixtures / slice / pm-mixture; (p, rho) = (0.383, 1.06|1.20) and (0.30, 2.0); 78 assembly runs, 80 budget-census fits; proofs (Lemmas A-C, Theorem P6', conditional theorem) are for all n but pending skeptic pass", + "refuted_by": null, + "gaps": [ + "averaged-odds-ratio-control", + "mutual-information-tax", + "recipe-totality", + "sinkhorn-tv-stability" + ], + "leak_terms": [ + "perturbative assembly", + "capped delta-ball", + "downward budget", + "centered-overlap", + "zero-prefix embedded crash", + "n-uniform budget", + "uc_pert", + "Theorem P6", + "tau_half", + "budget census" + ], + "note": "Sub-results at different confidence levels inside one record: the Prop-6 closure is a written-out proof (VERIFIED pending independent skeptic pass); the lead-3 assembly check and the non-uniformity of the pointwise budget are EVIDENCE at n <= 9; the assembly theorem is CONDITIONAL on budgets (B1)-(B4) with n-uniform constants, of which (B1) is the perturbative form of gap a-prime and (B2) of gap b. 005's leads 3 and 4 are distinct (the task line conflated them); both were run." + }, + { + "id": "009", + "file": "attempts/009-mutual-information-tax.md", + "date": "2026-07-31", + "mode": "informed", + "status": "EVIDENCE", + "mechanism": [ + "coupling", + "overlap-tilt", + "sinkhorn", + "chain-rule-assembly", + "mutual-information-tax" + ], + "one_line": "Gap 2 made precise as a chain-rule assembly statement (TAX at p) and probed: it survives all four adversary genres, is smoothing-sensitive on 002's own certificates, the pure slice tilt is provably tax-free, the half-mixing coupling provably loses nothing to chain-rule certification, and a previously unnamed second tax (measured ~0.40 log2 n on slices) is identified as the binding loss channel.", + "range": "Crash family n<=11, lambda in [-1,3]; half-mixing genre n<=14 at two parameter points; Sawin certificate gadgets n<=60000 (block recipe, rigorous bound); CL pure slice n<=80, lambda in [0,3], p in {0.42, 0.3823}. Finite-n, finite-grid EVIDENCE only; mini-lemmas SL/HM proved but not yet skeptic-reviewed.", + "refuted_by": null, + "gaps": [ + "averaged-odds-ratio-control", + "mutual-information-tax", + "recipe-totality", + "second-tax-scaling" + ], + "leak_terms": [ + "chain-rule assembly value", + "second tax", + "(TAX at p)", + "tax-free slice", + "uc_mitax", + "0.40 log2 n", + "lambda-window narrowing", + "CR = Gain identically" + ], + "note": "Deliberately the smallest-budget line of the cycle: a probe, not a campaign. The expected outcome was a precise obstruction; instead the candidate survived everywhere tested and the losses were localized (T_A vs ST, by genre)." + }, + { + "id": "010", + "file": "attempts/010-ideation-sweep-fields.md", + "date": "2026-07-31", + "mode": "informed", + "status": "MAP", + "mechanism": [ + "field-lens-sweep", + "barrier-analysis" + ], + "one_line": "First ideation sweep (6 field lenses): 3 queue-worthy routes (pairwise-closure LP certification, union-transfer-operator eigenvalue field, bipartite MIS decomposition), 6 secondary, 5 no-purchase verdicts incl. a proven product-weight no-go killing the multiplicative toolbox.", + "leak_terms": [ + "field-lens sweep", + "coatom eigenvalue", + "union transfer operator", + "eigenvalue field", + "log-supermodular lambda", + "product-weight no-go", + "pairwise-closure LP", + "union-injection Hall", + "union-escape graph", + "MIS share r(G)", + "crosscut truncation" + ], + "refuted_by": null, + "gaps": [] + }, + { + "id": "011", + "file": "attempts/011-skeptic-review-of-009.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED_WITH_CORRECTIONS", + "verifies": "009", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "coupling", + "overlap-tilt", + "sinkhorn", + "chain-rule-assembly", + "mutual-information-tax" + ], + "one_line": "Independent re-implementation (recursion engine, natural-log Sinkhorn, direct-MI cross-checks) confirms 009 in full: mini-lemmas SL and HM re-derived by hand and now VERIFIED, all headline numbers reproduce to <= 1.4e-10, the crash lambda-window sharpens to the law lam_max*(n-3) -> 4.8475, the ST ~ 0.40 log2 n fit survives extension to n = 240 and two new p, and the untested tilt-on-Sawin cell closes positive at n <= 300; corrections are reporting-level (grid-edge suprema quoted as suprema, mixed-p 'increasing margin', slice-ratio scope, a measured 1e-5 lgamma floor behind part D's 'exactly', and an unstated m_k = 1/2 legality hypothesis p_k >= 1-1/sqrt(2) in the block bound, verified to hold on every instance used).", + "corrections": [ + "slice sup_lambda CR values at p ~ 0.42 are grid-edge values; true suprema sit at lambda ~ 4-5 and are larger (e.g. +1.4655 vs reported +1.0199 at n=80, w0=34); 'increasing margin in n' conflates p = 0.4167 and p = 0.425 rows -- at fixed p the margin is cleanly increasing", + "Outcome scope 'p in {0.42, 0.3823}' overstates: slice ratios actually run are {0.4167, 0.425, 0.3833, 0.3875}; p = 0.3823 itself was never tested", + "part D's 'H(mu) computed exactly' has a measured ~4e-6 absolute lgamma floor at n = 60000 (harmless at the +2418 scale)", + "the part-D block bound's m_k = 1/2 choice needs the unstated hypothesis p_k >= 1-1/sqrt(2) ~ 0.29289; holds on all instances used but would fail silently on mixtures with a component marginal below it" + ], + "refuted_by": null, + "gaps": [ + "averaged-odds-ratio-control", + "mutual-information-tax", + "recipe-totality", + "second-tax-scaling" + ], + "leak_terms": [ + "uc_mitax_skeptic", + "mitaxsk", + "lambda-window law", + "4.847/(n-3)", + "0.405 log2 n", + "tilt-CR on Sawin", + "recursion engine", + "most conservative assembly", + "m_k legality" + ] + }, + { + "id": "012", + "file": "attempts/012-skeptic-review-of-008.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED_WITH_CORRECTIONS", + "verifies": "008", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "coupling", + "overlap-tilt", + "sinkhorn", + "perturbative-expansion", + "implicit-function-theorem" + ], + "one_line": "Line-by-line re-derivation and two independent engines (incl. an orbit-symmetrized census that reaches n = 32) confirm 008's Theorem P6' closure of the Prop-6 smoothness gap, the 78-run assembly table, the calibration block, and the centered-kernel closed form; but the conditional theorem's constants fail (box-uniform tau_half is analytically false at 008's own s0 = 0.10, ceiling s0 <= 0.0843; delta_0 = 0.022 should be <= 0.0079 by 008's own formula, ~0.004 corrected), and the extended census reverses the flattening reading: the averaged downward OR budget's increments stop shrinking and start growing past n ~ 22 (claimed limits 0.07/0.3 passed by n = 16/10, delta^2-coefficient verified stable in delta at n = 32), killing budget (B1) as stated at evidence level by 008's own lead-1 criterion, while the tax budget (B2) is first-order in delta, not quadratic, in every observable window.", + "corrections": [ + "conditional theorem's 'G nondecreasing in tau up to tau_half on the box' is analytically false at (p, rho, s0) = (0.383, 1.2, 0.10): (x0+s0)^2 = 0.5141 > 1/2 and z > xy for rho > 1, so tau_half(corner) = 0; the claimed P2d numerical check does not exist in uc_pert.py; repair requires s0 <= x_c - x0 = 0.0843 with re-measured constants", + "delta_0 ~ 0.022 does not follow from 008's stated formula: the code drops the ||G||_box/s0^2 c3 term (||G||_box = 0.1456, never computed by 008 -> delta_0 = 0.0079); the off-box charge needs the global sup(-G) = 0.487, not ||G||_box; M_tau must include the tau < 0 side (0.0838 vs 0.0592); c3 is ~4.7 not 2.5; fully corrected delta_0 ~ 0.004", + "the averaged downward budget does NOT flatten: at (0.383, 1.2) the delta->0 coefficient rises 0.065 (n=9) -> 0.104 (n=32) with increasing increments; at (0.30, 1.0) it passes the claimed 0.3 limit at n = 10 and reaches >= 1.24 by n = 32 (the fixed-delta turnover is a delta-saturation artifact); (B1) with n-uniform constant is dead at EVIDENCE level per 008's own lead-1 falsification criterion", + "the tax is first-order in delta (halving ratios -> 2), not O(delta^2), in every observable window at large n; fixed-delta tax/delta^2 exceeds the instantiated c2 = 0.01 by 10x at n = 32; (B2) must be restated as a linear-in-delta budget (absolute tax stays tiny, so end-to-end positivity is not threatened)", + "'worst in-class net = +0.0003 (slice n=5 delta=0.2)' is out of class by 008's own capped flag; true worst in-class min-net is ~ +0.00105 (rho=1.06 crash n=6 delta=0.02)", + "P6' constant chase: the stated step 'using m0 <= 1/2' yields 831, not 924 (924 corresponds to m0 <= 1; bound valid either way); the 'measured c ~ 0.6' looseness illustration ignores 008's own census worst case c ~ 20.6" + ], + "range": "Hand re-derivation of Lemmas A-C and Theorem P6' in full; P6' bound stress-tested on 120 instances (n <= 6, five directions incl. delta_full and half-empty, delta up to the theorem boundary 0.5, zero violations); Lemma B checked at interior delta; 78-run table spot-verified to 5e-6 and 104-run capped-negative hunt (rho in {1.045, 1.06, 1.20}, n <= 9, delta <= 0.2, incl. the half-empty direction with marginals exactly p) found none; budget census re-run and extended to n = 32 on the crash direction via an orbit-symmetrized engine (n <= 9 values reproduce 008 to all printed digits), with delta -> 0 stability verified at n = 32 down to delta = 0.003125 for the downward budget (clean delta^2) and shown NOT to hold for the tax (delta-linear); conditional-theorem constants recomputed on 008's grid plus tau < 0 and global domains", + "gaps": [ + "averaged-odds-ratio-control", + "mutual-information-tax", + "recipe-totality", + "sinkhorn-tv-stability" + ], + "leak_terms": [ + "uc_pert_skeptic", + "pertsk", + "orbit engine", + "budget growth resumes", + "delta-linear tax", + "s0 ceiling 0.0843", + "tau_half corner", + "corrected delta0", + "n = 32 census" + ] + }, + { + "id": "013", + "file": "attempts/013-skeptic-review-of-007.md", + "date": "2026-07-31", + "mode": "informed", + "status": "VERIFIED", + "verifies": "007", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "exact-rational-arithmetic", + "coupling", + "overlap-tilt", + "sinkhorn", + "gram-mass-pairing" + ], + "one_line": "Exact rational arithmetic (Fraction end-to-end, certified log2 enclosures) confirms 007 in full: the 10-atom witness's violation is certified float-free at rational tilts spanning lambda ~ 0.0144..5 including exactly rational lambda = 2, 3, 4 (positive at small t exactly where Theorem C forces it), both invariances hold as exact rational identities, the aggregate is certified positive (+1.844669) and survives witness-seeded two-coordinate attacks, all four partial theorems re-derive by hand, and the refutation even survives the conjecture-friendliest alternative bookkeeping (degenerate histories scored at lambda: still -0.067); no corrections, one cosmetic runtime discrepancy.", + "range": "Witness certified at 12 exact rational tilt parameters (fixed-u family t = 101/100 .. 32 plus rationalized-refit measures at lambda = 1, 2, 3 within 9.2e-16 of the witness mu); enclosure widths < 1e-40; own 96-instance battery + 21-endpoint aggregate hunt at n <= 7; Theorem B probed on 500 random <= 4-atom instances; hand re-derivations of section 1, Lemmas G/M, Theorems A/B/C and both invariances", + "refuted_by": null, + "gaps": [ + "aggregated-or-control", + "margin-modulated-or-control", + "mutual-information-tax", + "recipe-totality" + ], + "leak_terms": [ + "oravgsk", + "uc_or_avg_skeptic", + "exact rational witness", + "convention robustness", + "surrogate-lambda", + "1.844669", + "minor ratio 6.68" + ], + "note": "Decisive check: 007's confessed vulnerability (both engines share IEEE floats; its first witnesses were shared-underflow artifacts) is closed by re-verifying the witness in exact rational arithmetic with directed-rounding log2 enclosures. At t = 4 and t = 16 the refutation of M_i >= lambda is a fully rational statement with lambda = 2, 4 exactly. New strengthening beyond 007: the violation does not hinge on the conditioning-out convention (degenerate histories scored at exactly lambda still give M_5 - lambda = -0.067, certified). Nuance: exact certificates at lambda = 1, 2, 3 are for rationalized-refit measures within 9.2e-16 of the witness mu, not the identical rational mu. Theorem C's smoothness step is analogous to but not literally discharged by 008's P6-prime (P6-prime is d/d-mu at fixed lambda; Theorem C needs d/d-lambda at fixed mu)." + }, + { + "id": "014", + "file": "attempts/014-aggregated-control-probe-large-n.md", + "date": "2026-08-04", + "mode": "informed", + "status": "REFUTED", + "mechanism": [ + "coupling", + "overlap-tilt", + "sinkhorn", + "plackett", + "gram-mass-pairing", + "exact-rational-arithmetic" + ], + "one_line": "Large-n probe of the restated Gap 1 kills it: unit-replicating the 10-atom witness genre (r crash units sharing one block marker) with re-optimized shared weights drives the i-aggregated control negative past n ~ 90, certified float-free at t = 4 (n = 96 and 160, marginals <= 0.309); the n <= 32 space stays positive everywhere searched.", + "leak_terms": [ + "unit replication", + "multi-unit witness ladder", + "MU(n,r)", + "theta-star ladder", + "aggregate crossing", + "O(1) surplus corridor", + "0.024226", + "-0.000759865", + "uc_or_agg_probe", + "aggprobe" + ], + "range": "Free-support seeded climbs n <= 32 (58 climbs round 1, deep annealed round 2) plus permutation orbits: 0 violations; unit-symmetric ladder evaluated to n = 160 (raw witness weights: positive through n = 128, min +0.157; optimized theta*: float crossing in (88, 96], -0.0242 at n = 160, 20/20 perturbations stay negative); exact rational certificates at t = 181/16 and t = 4, enclosure widths < 1e-18", + "refuted_by": null, + "gaps": [ + "margin-modulated-or-control", + "mutual-information-tax", + "recipe-totality" + ], + "note": "Settles queue item 1's probe-first branch negatively: the aggregated control is certified positive at n <= 32 on everything searched, so proof effort there was exactly the small-n trap 012 warned about. The kill is convention-robust (scoring degenerate histories at lambda only grows the denominator, never the numerator). Route stays LIVE; Gap 1 falls to the margin-modulated candidate (007 lead 2), which this record's ladder family must now be run against." + }, + { + "id": "015", + "file": "attempts/015-skeptic-review-of-014.md", + "date": "2026-08-04", + "mode": "informed", + "status": "VERIFIED_WITH_CORRECTIONS", + "verifies": "014", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "exact-rational-arithmetic", + "coupling", + "overlap-tilt", + "sinkhorn", + "gram-mass-pairing" + ], + "one_line": "From-scratch third implementation of the aggregate (scaled-integer census plus an interval-squaring log2 enclosure, nothing shared with the probe engines) confirms 014's kill: all three certificates re-derive with matching digits (n=96 in [-0.000759865185844166092, -0.000759865185844166079], n=128 -0.0154057269, n=160 -0.0242259093, all t=4, marginals certified below 0.38271, dichotomy exact), the anchors and positive controls re-certify (+1.844669005 witness; raw-weight ladder at n=96 upgraded to certified +0.157131854664), and the convention-robustness proof re-derives; corrections: the n=128 instance has 252 atoms not 253, and P4's claim that the violating lambda-window sits where 009/011's lambda-window law puts the workable range is wrong (at n=96 the law gives lambda_max ~ 0.052 vs the violation at lambda in [2, 2.5]), so the lambda-restricted restatement joins the margin-modulated control as a live Gap-1 candidate.", + "corrections": [ + "n = 128 tidy certificate instance has 252 atoms (2r+8, r=122), not the 253 stated in 014's certification block and Outcome table", + "P4's 'violating window sits exactly at moderate lambda where 009/011's lambda-window law puts the workable range' is false at every violating n: lambda_max ~ 4.847/(n-3) ~ 0.052 at n = 96 while the violation lives at lambda in [2, 2.5] and the same family is positive at lambda <= 1.5 -- the kill closes the unrestricted for-all-lambda gap only, and the lambda-restricted (lambda <~ c/n) restatement is untouched by this family" + ], + "range": "Kill certificates independently re-derived at n = 96 (188 atoms), 128 (252 atoms), 160 (316 atoms), all at t = 4 exactly, enclosure widths <= 1.3e-20; witness anchor at t = 181/16; P3-best n = 32 and raw-witness ladder n = 96 positives certified; padding invariance MU(20,1)/MU(32,1) verified as exact rational identity; alternative-bookkeeping aggregate certified negative at all three n; MU builder cross-checked atom-for-atom against the probe's; 014's float-only side claims (crossing in (88,96], 20/20 perturbations, P4 window edges) reconciled against checkpoints but not certified; n <= 32 search completeness not re-run (stays EVIDENCE)", + "refuted_by": null, + "gaps": [ + "margin-modulated-or-control", + "lambda-restricted-or-control", + "mutual-information-tax", + "recipe-totality" + ], + "leak_terms": [ + "aggsk15", + "uc_or_agg_skeptic", + "interval-squaring enclosure", + "0.157131854664", + "lambda-window disjointness", + "252 atoms not 253" + ], + "note": "Decisive check per the 2026-07-31 verification standard: 014's certificates reused 013's exact engine and enclosures, so this review removes the shared-code risk with a third census written from 007 section 1's prose and a structurally different certified log2 enclosure (dyadic interval squaring; ln 2 from the exact series sum 1/(k 2^k)). Free anchors verified exactly: OR/t = 1 at i = n (005 Prop 1) on every instance, and including the i = n term keeps every kill negative. Union-closure of the support is correctly NOT a hypothesis (the control quantifies over all mu; 007's witness was not union-closed either) -- checked and recorded so the scope is explicit. Precedent followed from 013/007: the reviewed entry (014) is not modified." + } + ] +} From a6620f181f46c3d3515c1aa8e9d87b01da5ff8cd Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 11:00:19 +0000 Subject: [PATCH 11/14] Ledger: integrate the second 2026-08-04 cycle (billiards + union-closed lines) TL;DR rewritten for the cycle headline (aggregated OR control refuted at large n, Laurent block formalized); union-closed and billiards problem rows updated; queue item 1 restated to the two surviving Gap-1 candidates with probe-first designs, old item 15 replaced by the kill-frontier follow-ups; verified-results entries added for the Laurent block (011/012) and the aggregated-control refutation (014/015); new dead-end entry for the unrestricted-lambda aggregated form with the ladder genre as reusable adversary; unit-replication insight recorded; GUIDANCE and FORMALIZE.md pilot list updated to completed state. Crouzeix blind line still pending integration. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- GUIDANCE.md | 7 +-- STATUS.md | 114 +++++++++++++++++++++++++++++++++++++--------- docs/FORMALIZE.md | 5 +- 3 files changed, 100 insertions(+), 26 deletions(-) diff --git a/GUIDANCE.md b/GUIDANCE.md index 735a583..b89d710 100644 --- a/GUIDANCE.md +++ b/GUIDANCE.md @@ -33,9 +33,10 @@ what matters here, and three things follow for this lab: independent derivation — it is what made Astra's claims land after the October episode. `docs/FORMALIZE.md` defines the lane and the `FORMALIZED` status; the pilot completed 2026-08-04 (billiards 005's - Lemma L1, attempts 009+010, full three-part pass). Next certificate - target is the I1–I4 Laurent block (queue item 15). Expensive; use - deliberately. + Lemma L1, attempts 009+010) and the lane scaled the same day to the + full I1–I4 Laurent block (011+012, 23 theorems, zero corrections). + The remaining named candidate is union-closed P6′ (`docs/FORMALIZE.md` + pilot list). Expensive; use deliberately. **Portfolio note.** Astra's target list — decade-open but specialist-tractable problems whose deliverables are checkable objects diff --git a/STATUS.md b/STATUS.md index 994d36c..46013c9 100644 --- a/STATUS.md +++ b/STATUS.md @@ -11,16 +11,26 @@ out, and no cron trigger is active. Nothing is half-finished: every line is written up, and the queue below is a list of starting points rather than abandoned work. -**2026-08-04 cycle (Astra follow-through):** a literature-novelty check is -now part of the skeptic pass and a formalization lane (`docs/FORMALIZE.md`, -status `FORMALIZED`) sits above `VERIFIED` for load-bearing proof steps. -Both queue items the new guidance created are done: the ripple scan of the -ten Astra results ran with **zero hits** across the portfolio (see -insights — both flagged bites are reasoned misses), and the formalization -pilot landed: billiards 005's Lemma L1 is the lab's first machine-checked -result (Lean 4 + mathlib, attempts 009+010, statement review and -independent rebuild complete). No mathematical state changed beyond the -new certificate. +**2026-08-04, two cycles.** First (Astra follow-through): a +literature-novelty check joined the skeptic pass, the formalization lane +(`docs/FORMALIZE.md`, status `FORMALIZED`) went in above `VERIFIED`, the +ripple scan of the ten Astra results ran with **zero hits** (see +insights), and the pilot certificate landed (billiards L1, 009/010). +Second cycle HEADLINE: **the i-AGGREGATED odds-ratio control — Gap 1's +restatement — is REFUTED at large n** (014, skeptic-confirmed 015): the +10-atom witness replicates into a ladder MU(n,r) whose aggregate falls +like a − b·log n and crosses zero near n ≈ 90, certified in exact +rational arithmetic at n = 96/128/160 (λ = 2, marginals ≤ 0.309), the +skeptic's from-scratch engine matching to 18+ digits. The kill closes +the recorded ∀λ gap; the surviving Gap-1 candidates are the +margin-modulated control and the λ ≲ c/n window-restricted variant (the +violation lives at λ ∈ [2, 2.5], ~40× above the workable window at +n = 96). Probe-before-proof vindicated a second time: certified +positives at n ≤ 32 masked the asymptotic failure. Also this cycle: the +**entire 005 Laurent block is now FORMALIZED** (011/012 — closed-form +composition, I1–I4, glide facts, specializations; 23 theorems, zero +corrections, statement review + independent rebuild complete), and the +first blind attempt on crouzeix ran (integration in progress). To pick something up, take an item from the attempt queue and follow `docs/CYCLE.md` — either by hand, one line at a time, or by restarting an @@ -28,7 +38,7 @@ hourly routine with that file as the prompt. Everything needed is in this directory. New attempts are welcome as pull requests; see `CONTRIBUTING.md`, and `AGENTS.md` for whether to work blind or informed. -Current standing. HEADLINE: the second 2026-07-31 cycle ran the +Standing from the 2026-07-31 cycles: the second cycle ran the two-track billiards plan (old queue 11+12); all four records are skeptic-confirmed, six corrections total, none load-bearing. Conservative track (005+008): the W(a,b) death law is closed on both @@ -56,28 +66,28 @@ surviving; route LIVE, ceiling 0.4315 vs record 0.38271. Standing library: Singmaster census to 2.5×10^29; Erdős–Gyárfás cubics to n=22 + girth-≥5 at n=24; lonely-runner k=8; Erdős–Straus identity-poverty; graceful n ≤ 14; mahler-4d and billiards blind -censuses. 21 verified results, 7 recorded dead ends, 10 problems, -~45 reusable tools. Crouzeix remains the one problem with no -attempts (queue; run blind). +censuses. As of 2026-08-04: 24 verified results, 8 recorded dead +ends, 10 problems, ~48 reusable tools, and a machine-checked +certificate layer (24 Lean theorems) over the billiards route. ## Problem status | Problem | Status | Budget | Active line | |---|---|---|---| | Erdős–Gyárfás | n=24 done | high | next: 2-connected C16-free test (lead 2); n=26 is a ~16h run | -| Union-closed (Frankl) | ROUTE LIVE (verified) | high | gap (a′) refuted (007/013) → live gaps: aggregated OR control (probe n ≳ 20 BEFORE proof effort), δ-linear tax budget, corrected assembly budgets (012); plus three sweep leads (010) | +| Union-closed (Frankl) | ROUTE LIVE (verified) | high | gap (a′) refuted (007/013); AGGREGATED control refuted at large n (014/015, certified n = 96–160) → live: margin-modulated control, λ ≲ c/n restricted variant, δ-linear tax budget, corrected assembly budgets (012); plus three sweep leads (010) | | Erdős–Straus | 601 resolved | medium | next: prove the identity-poor mechanism (why QR classes force low f) | | Singmaster | census done | medium | next: Diophantine curve table (search-deeper is now low value) | | Lonely runner | k=8 done | medium | next: k=9 scan (k=8 likely settled by Rosenfeld preprint) | | Graceful trees | census done (n≤14) | low | possible next: mine the symmetric-spider seed; lobster verification at larger n | | Collatz | queued | low (long shot) | failed-approach taxonomy; cycle-bound frontier | -| Triangular billiards | death law CLOSED both sides: parametric necessity all (a,b), death = γ_d exactly for 20 members (005/008); 135° stall dissolved, birth law + exact-135 certificates (006/007) | high | next: parametric sufficiency + birth-law theorem (queue 11); coverage conjecture + sampler blind spot (queue 12); Lean lane open — L1 FORMALIZED (009/010), I1–I4 Laurent block is the next certificate target | +| Triangular billiards | death law CLOSED both sides: parametric necessity all (a,b), death = γ_d exactly for 20 members (005/008); 135° stall dissolved, birth law + exact-135 certificates (006/007) | high | next: parametric sufficiency + birth-law theorem (queue 11); coverage conjecture + sampler blind spot (queue 12); Lean lane: L1 + the full Laurent block FORMALIZED (009–012); the geometry bridge stays informal by design | | Mahler in ℝ⁴ | census done blind (skeptic-confirmed) | medium | next: close k=12–20 (falsifiable: no proper mask with P<11); run the same pipeline on {0,±1}³ for the n=3 spectrum comparison | | Crouzeix | onboarded, no attempts | medium | harness ready, certification risk retired; first attempt is the dim-3 landscape (run blind) | ## Attempt queue (next cycles pull from the top) -1. [union-closed] The restated gap: i-AGGREGATED odds-ratio control (007's replacement after the per-i averaged form died — see 007/013). FIRST extend the certified probe to larger n: 012's orbit-symmetrized engine reaches n = 32, and its budget-growth reversal past n ≈ 22 is a standing warning against small-n optimism — the aggregate is certified positive only at n ≤ 7. If it survives n ≳ 20, attack the proof through 007's Gram/Frobenius machinery (PSD mass matrix; first-order-in-λ perfect square) and the margin-modulated variant. Any claimed bound → skeptic review before ledger entry. +1. [union-closed] The surviving Gap-1 candidates after 014/015's kill of the aggregated control, probe-first per standing policy: (i) **margin-modulated OR control** — run the certified probe on the MU(n,r) ladder genre FIRST (it is the reusable adversary; 015's independent engine `uc_or_agg_skeptic.py` certifies), then the killer genres from 007/013; only if it survives n ≳ 96 does proof effort start (007's Gram/Frobenius machinery). (ii) **λ-restricted variant**: the control restricted to the workable window λ ≲ 4.847/(n−3) (the 009/011 λ-window law) — 014's violation lives at λ ∈ [2, 2.5] and its family is positive at λ ≤ 1.5, so this variant is genuinely untested; probe the ladder AT the window boundary across n before anything else. Any claimed bound → skeptic review before ledger entry. 2. [union-closed] Rebuild the assembly budgets per 012's corrections: the tax is δ-LINEAR (restate B2), the τ_half step needs s₀ ≤ 0.0843, corrected δ₀ ≈ 0.004. The open question is whether ANY n-uniform budget object exists — 008's conditional theorem survives structurally; its constants need the corrected inputs, and the budget census machinery (uc_pert.py + uc_pert_skeptic.py's orbit engine) is ready. 3. [union-closed] Sweep leads from 010, one per cycle: (i) pairwise-closure LP/degree-2 SOS certification on the n ≤ 4 census (all 4958 families) — kill: certified bound converges to ≈ 0.382; (ii) union-transfer-operator eigenvalue field — test log-supermodularity of λ_C on the n ≤ 5 census (pre-check the total-positivity records first); (iii) bipartite-MIS decomposition — geng census to n = 12, connectivity of extremals, compositionality across 1- and 2-cuts. Kill conditions recorded in 010. 4. [erdos-straus] Prove the identity-poverty mechanism: why does QR-class membership mod 840 force fewer Type I covering congruences? Start from 001's obstruction analysis + 002's rate data; target a theorem "f(p) ≥ g(N_typeI(p))" or a disproof. @@ -91,7 +101,7 @@ attempts (queue; run blind). 12. [billiards-triangles] **The coverage conjecture, and the sampler blind spot** (from 006/007, 2026-07-31; absorbs the old pinch-gap item — its motivating gap [135.000°, 135.049°] is CLOSED, W(4,3) is certified alive inside it): 006 reduced "every obtuse angle has an alive W member" to an elementary Diophantine statement (unproven; float-checked at 157 + 25 arcs over 90.5°–165°, zero failures). Prove it, using the birth law as a labelled input where needed. Note the certificates so far are POINTWISE (007's C2): window-interval continuity on sub-arcs is float + SPECULATION law only, and per-triangle coverage of a whole arc is a different (open) question — the windows are x-slivers at the corners. Separately falsifiable (007 lead): every sampler in use accumulates only at the 90/j window edges, so an interior-pinch alive window would hide from ALL current designs — build one targeted interior-accumulation test before trusting any negative screen again. 13. [mahler-4d] Close the {0,±1}⁴ universe: k = 12–20 pairs (~30M orbits at k=12, improper fraction already 77% at k=9). Falsifiable: no proper mask with k ≥ 12 has P < 11. Needs the improper-detection shortcut or a streaming canonicalizer; see 001 lead 1. Cheap side quest, same pipeline: the {0,±1}³ census for the n=3 spectrum comparison (13 pairs, trivial) — does the non-Hanner gap grow or shrink with n? 14. [billiards-triangles] Coverage self-test: re-derive the acute and right-triangle cases as a scoped attempt record. Low value now that the harness self-test covers Fagnano and the orthic geometry and 001 mapped the obtuse side — take it only if something turns up that the certificate machinery cannot express. -15. [billiards-triangles] **Second certificate: the I1–I4 Laurent-identity block** (`docs/FORMALIZE.md`; follows the completed L1 pilot, 009/010). The identities are already one exact polynomial check after adjoining e^{iaα}, e^{ibβ} as formal variables — formalizing the polynomial identity is the tractable half, and the geometry bridge (unfolding = closed form) is exactly what the statement-review step exists to interrogate; 010 kept this gap open deliberately. Harder than the pilot: needs Laurent-polynomial infrastructure in mathlib, and the statement must be pinned to what 005 actually proves (formal identity + ring-homomorphism specialization), not to the geometric claim it feeds. +15. [union-closed] **Push the kill's frontier** (cheap, from 014 leads): direct θ-optimization of the MU ladder at target n to find the minimal violating n (currently bracketed (32, 96]); extend the raw-weight ladder past n = 128 (certified +0.157 at 96 — does it also eventually cross?); and re-run the 014 ladder at the λ-window boundary as the first data point for queue item 1(ii). ## Verified results @@ -334,12 +344,59 @@ attempts (queue; run blind). StrictAntiOn, and Real.cot semantics read from mathlib source), the cheat scan is clean, the root module provably elaborates the theorem file, and the reviewer's own rebuild + axiom audit went green. Scope: - exactly the formal statement — L1 only; Lemmas C/D themselves and the - I1–I4 Laurent block remain VERIFIED, not FORMALIZED (queue 15). Source - in problems/billiards-triangles/formal/. + exactly the formal statement — L1 only; the I1–I4 Laurent block + followed the same day (011/012, next entry). Source in + problems/billiards-triangles/formal/. +- **[billiards-triangles] The full 005 Laurent block FORMALIZED + (skeptic-confirmed)** (2026-08-04, attempts 011+012): the closed-form + half-word composition, pair collapses, u² = translation, glide action, + identities I1–I4/D1/D2, and the ℤ-specialization corollaries — 23 + theorems in Lean 4 (mathlib v4.32.2), zero sorries, axioms exactly + {propext, Classical.choice, Quot.sound}, with (a,b) carried as actual + powers of ∀-quantified generators so one check covers all integers. + Statement review (012): all 23 statements mapped to 005 verbatim + (angle monomials, factor-of-2 conventions, signs; permutation/sign + hunts empty), conjugation-as-substitution confirmed as 008's star + involution with nothing smuggled, the dropped i² = −1 hypothesis + independently re-proved safe (the block is homogeneous in i), 31/31 + identities re-derived exactly in a from-scratch stdlib Laurent engine, + independent rebuild + axiom audit green. Method notes: no Laurent-ring + mathlib API needed (arbitrary field + field_simp + ring_nf) — the + anticipated obstruction never materialized. Scope: the ring-model + statements only; the geometry bridge (unfolding = closed form) is + permanently carried by 005/008's informal cross-checks. + +- **[union-closed] i-AGGREGATED odds-ratio control REFUTED at large n + (skeptic-confirmed)** (2026-08-04, attempts 014+015): the control — + certified positive at n ≤ 7 by 013 and surviving 58 seeded attacks at + n ≤ 32 — fails asymptotically: replicating the 10-atom witness's + crash core Θ(n) times with orbit-optimized shared unit weights gives + a ladder MU(n,r) whose aggregate falls like a − b·log n, certified + negative in exact rational arithmetic at n = 96 (−7.6e-4, enclosure + width ≤ 1.3e-20), n = 128 (−0.0154), n = 160 (−0.0242) at t = 4 + (λ = 2 exactly), all marginals ≤ 0.309 < 0.38271, sign robust to the + alternative degenerate-history bookkeeping. 015 re-derived the + quantity from 007/013's prose, re-implemented everything from scratch + (own log₂ enclosure via interval squaring), rebuilt the family + atom-for-atom, and matched all three certificates to 18+ digits; + positive controls confirmed, the surviving raw-weight ladder upgraded + to certified +0.157 at n = 96. Corrections: n = 128 atom count (252); + and the violation lives at λ ∈ [2, 2.5], NOT in the 009/011 workable + window (≈ 0.05 at n = 96) — so the ∀λ gap closes but the + window-restricted variant is untouched (queue 1). ## Insights / cross-problem notes +- Unit replication + re-weighting is a portable adversary genre + (2026-08-04, union-closed 014/015): a family certified positive at + small n can fail after replicating its adversarial core Θ(n) times + and re-optimizing shared weights — the zero crossing near n ≈ 90 was + invisible to every fixed-n search below it. Second vindication of + probe-before-proof. Companion lesson from 015's correction C2: a kill + must be checked against the workable parameter window before + declaring a branch fully closed — 014's kill closes the recorded ∀λ + statement; the window-restricted variant needed its own verdict and + didn't get one from this family. - Ripple scan of the Astra results ran (2026-08-04, informed): all ten results characterized (press-transcribed only — openai.com, the certificate repo, and erdosproblems.com were unreachable, so every scope @@ -527,7 +584,20 @@ attempts (queue; run blind). convention. Do not re-attempt any per-i averaged form; the witness genre (near-proportional light slices, anti-aligned cross-ratios) is the reusable adversary. Live replacements: the i-aggregated and - margin-modulated controls (queue item 1). + margin-modulated controls (queue item 1). [Update 2026-08-04: the + i-aggregated replacement is now also dead — see below.] +- **[union-closed] i-AGGREGATED odds-ratio control, unrestricted-λ form + (013's restatement)** (2026-08-04, attempts 014+015): dead at large n — + the replicated-witness ladder MU(n,r) with orbit-optimized shared unit + weights is certified negative in exact rational arithmetic at + n = 96/128/160 (λ = 2, marginals ≤ 0.309), independently re-derived to + 18+ digits by 015's from-scratch engine. Do not re-attempt any ∀λ + aggregated form; the ladder + re-weighting genre is the reusable + adversary (`explore/uc_or_agg_probe2.py` builds it, + `explore/uc_or_agg_skeptic.py` certifies independently). NOT covered + by the kill: the margin-modulated control and the λ ≲ c/n + window-restricted variant (the violation sits at λ ∈ [2, 2.5], far + above the workable window) — both live, queue item 1. - **[union-closed] n-uniform δ²-budgets for the perturbative assembly (008's B1/B2 as stated)** (2026-07-31, attempt 012): the averaged downward budget's flattening reverses past n ≈ 22 (evidence to diff --git a/docs/FORMALIZE.md b/docs/FORMALIZE.md index 6241367..79874f0 100644 --- a/docs/FORMALIZE.md +++ b/docs/FORMALIZE.md @@ -76,7 +76,10 @@ So a `FORMALIZED` pass has three parts, and the first is the point: ## Pilot candidates (2026-08-04) -Ordered by (value ÷ effort), from the current ledger: +Ordered by (value ÷ effort), from the current ledger. [Status, end of +2026-08-04: (1) and (2) are DONE — billiards attempts 009–012 formalized +the cot lemma and then the full Laurent block including I1–I4, both with +completed statement reviews; (3) remains open.] 1. **Billiards Laurent-identity block** (005/008): I1–I4 and the glide facts are already "one exact polynomial check" after adjoining e^{iaα}, e^{ibβ} From 2e7de6830377cc7fd1265b7d823c0e0820ec0fb1 Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 11:07:34 +0000 Subject: [PATCH 12/14] crouzeix 001: blind dim-3 local-maxima census merged - skeptic pending First attempt on the last untouched problem, run blind per the queue (worked in a blind.sh copy; prior art physically absent). 286-start census of the Crouzeix ratio at n=3, deg<=3 from seven structured start families, certified rational enclosures at every endpoint, zero refutations: all probe-surviving local maxima are known structure (13 ice-cream-cone points at ratio 1 - a blind rediscovery of Greenbaum-Overton's configuration, cited; 97 Jordan-type endpoints climbing to ratio ~2); all four intermediate candidates fell to deep escape-probing as optimizer stalls. Standing gap recorded: the published GO intermediate maxima are untested by this design. Merge adds the four mechanism-tag definitions per the blind-flow rule. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- mechanisms.json | 22 +- .../001-blind-dim3-local-maxima-census.md | 253 +++++++++ problems/crouzeix/data/census_i4.jsonl | 79 +++ problems/crouzeix/data/census_i5.jsonl | 83 +++ problems/crouzeix/data/census_i6.jsonl | 79 +++ problems/crouzeix/data/census_w1.jsonl | 19 + problems/crouzeix/data/census_w2.jsonl | 13 + problems/crouzeix/data/census_w3.jsonl | 13 + problems/crouzeix/data/certified.jsonl | 286 ++++++++++ problems/crouzeix/data/certify.log | 70 +++ problems/crouzeix/data/classified.jsonl | 286 ++++++++++ problems/crouzeix/data/classify.log | 26 + problems/crouzeix/data/deep_probe.log | 4 + problems/crouzeix/explore/census.py | 505 ++++++++++++++++++ .../crouzeix/explore/certify_endpoints.py | 111 ++++ problems/crouzeix/explore/classify.py | 120 +++++ problems/crouzeix/explore/deep_probe.py | 32 ++ problems/crouzeix/explore/measure_cost.py | 18 + problems/crouzeix/explore/survivors_detail.py | 104 ++++ problems/crouzeix/prior-art.json | 34 +- 20 files changed, 2150 insertions(+), 7 deletions(-) create mode 100644 problems/crouzeix/attempts/001-blind-dim3-local-maxima-census.md create mode 100644 problems/crouzeix/data/census_i4.jsonl create mode 100644 problems/crouzeix/data/census_i5.jsonl create mode 100644 problems/crouzeix/data/census_i6.jsonl create mode 100644 problems/crouzeix/data/census_w1.jsonl create mode 100644 problems/crouzeix/data/census_w2.jsonl create mode 100644 problems/crouzeix/data/census_w3.jsonl create mode 100644 problems/crouzeix/data/certified.jsonl create mode 100644 problems/crouzeix/data/certify.log create mode 100644 problems/crouzeix/data/classified.jsonl create mode 100644 problems/crouzeix/data/classify.log create mode 100644 problems/crouzeix/data/deep_probe.log create mode 100644 problems/crouzeix/explore/census.py create mode 100644 problems/crouzeix/explore/certify_endpoints.py create mode 100644 problems/crouzeix/explore/classify.py create mode 100644 problems/crouzeix/explore/deep_probe.py create mode 100644 problems/crouzeix/explore/measure_cost.py create mode 100644 problems/crouzeix/explore/survivors_detail.py diff --git a/mechanisms.json b/mechanisms.json index 3cc92c2..bcca12e 100644 --- a/mechanisms.json +++ b/mechanisms.json @@ -1,6 +1,6 @@ { - "_comment": "The cross-problem approach taxonomy. Every `mechanism` tag used in any problems/*/prior-art.json must be defined here with a field and a one-line description; tests/test_mechanisms.py enforces the sync. Query it with scripts/mechanisms.py (list / where / gaps / matrix). Descriptions are METHOD-level, never findings \u2014 findings live in attempt records and their leak_terms. This file is tier 1: blind agents never see it.", - "_fields_comment": "`fields` is the lens vocabulary for cross-pollination. Each mechanism tag belongs to exactly one field. A field with no tags on a given problem is an untried lens there \u2014 that is what docs/IDEATE.md sweeps. Fields with zero tags anywhere are deliberately listed: they are the widest-open lenses.", + "_comment": "The cross-problem approach taxonomy. Every `mechanism` tag used in any problems/*/prior-art.json must be defined here with a field and a one-line description; tests/test_mechanisms.py enforces the sync. Query it with scripts/mechanisms.py (list / where / gaps / matrix). Descriptions are METHOD-level, never findings — findings live in attempt records and their leak_terms. This file is tier 1: blind agents never see it.", + "_fields_comment": "`fields` is the lens vocabulary for cross-pollination. Each mechanism tag belongs to exactly one field. A field with no tags on a given problem is an untried lens there — that is what docs/IDEATE.md sweeps. Fields with zero tags anywhere are deliberately listed: they are the widest-open lenses.", "fields": { "algebra": "Group actions, symmetry, invariants, polynomial and ring structure.", "analysis": "Real/complex analysis, generating functions, asymptotics, inequalities over the reals.", @@ -28,6 +28,10 @@ "field": "logic-methodology", "description": "Map where and why a known technique provably stops, to direct later attempts around the obstruction." }, + "basin-classification": { + "field": "computation", + "description": "Attribute each multi-start endpoint to a structure class via structural fingerprints and re-optimization probes, so landscape claims quantify basins rather than isolated values." + }, "bitmask-backtracking": { "field": "computation", "description": "Depth-first search over labelings/assignments with bitset state for constant-time constraint checks." @@ -52,6 +56,10 @@ "field": "computation", "description": "Cover a parameter region by boxes on which a claim is certified by exact or validated arithmetic, and treat the geometry of the uncovered remainder as a first-class deliverable." }, + "certified-enclosure": { + "field": "computation", + "description": "Turn a float optimization endpoint into a proven bound by rounding to a rational point and certifying the objective there with rigorous directed-rounding enclosures." + }, "certified-interval-inequality": { "field": "analysis", "description": "Prove a one-variable analytic inequality rigorously by splitting into an endpoint-algebra zone, an adaptive rational-interval bisection zone, and a derivative-sign zone near a vanishing endpoint; all enclosures exact-rational, single evaluations only." @@ -168,6 +176,14 @@ "field": "computation", "description": "Canonical isomorph-free graph generation via nauty/geng as the enumeration backbone." }, + "nonsmooth-maximization": { + "field": "analysis", + "description": "Locally maximize a nonsmooth (norm/eigenvalue/max-type) objective with derivative-free methods, using an explicit escape-probe test to separate genuine local maxima from optimizer stalls." + }, + "optimization-landscape-census": { + "field": "computation", + "description": "Map the local-maximum landscape of an objective by multi-start local optimization from designed start families, classifying every endpoint rather than reporting only the best value." + }, "out-of-sample-prediction": { "field": "logic-methodology", "description": "Test a conjectured law on family members it was not fitted to, computed before comparison, so that agreement is evidence and disagreement is refutation rather than refitting." @@ -253,4 +269,4 @@ "description": "Replace entropy differences by c-weighted KL divergences to a reference law and tune c." } } -} +} \ No newline at end of file diff --git a/problems/crouzeix/attempts/001-blind-dim3-local-maxima-census.md b/problems/crouzeix/attempts/001-blind-dim3-local-maxima-census.md new file mode 100644 index 0000000..65d3cd9 --- /dev/null +++ b/problems/crouzeix/attempts/001-blind-dim3-local-maxima-census.md @@ -0,0 +1,253 @@ +# Local-maxima census of the Crouzeix ratio, n = 3, deg p <= 3 + +``` +problem: crouzeix +date: 2026-08-04 +mode: blind +source-commit: a0bfd6ed660fb4ab8ced4158ba491c87a4667cd8 +type: computational (optimization-landscape census; certified enclosures) +tools: explore/census.py, explore/classify.py, explore/survivors_detail.py, + explore/certify_endpoints.py, explore/measure_cost.py (all pure + stdlib), harness/crouzeix/ratio.py, harness/crouzeix/verify_ratio.py +sources: PROBLEM.md; published literature via web search (Greenbaum-Overton + papers only; nothing about this lab) +``` + +## Approach + +Question: starting local maximization of the Crouzeix ratio +`f(A,p) = ||p(A)|| / max_{z in W(A)} |p(z)|` from a designed sample of +(matrix, polynomial) configurations at the fixed slice **n = 3, deg p <= 3**, +do all local maxima terminate at already-known extremal structure, or does +anything new appear? + +Why this: the conjecture is exactly the claim that the global maximum of f is +2. A census of *local* maxima at the smallest open dimension is the cheapest +way to (a) look for unexpected near-extremal structure and (b) produce +certified, reproducible landscape evidence, which the published numerical +work does not carry. + +**How this differs from the published Greenbaum–Overton experiments** (their +LAA 2018 paper and the Calcolo 2021 follow-up; see References — stated up +front because reproducing their finding is not a result): + +1. **Fixed slice.** GO sweep many dimensions and degrees; this census fixes + n = 3, deg <= 3 and spends the budget on basin coverage of that one slice. +2. **Structured start families.** GO initialize from random (2021: + heavy-tailed) matrices and polynomials. Here 7 named start families are + used, five of them deliberately placed at or near non-generic structure: + perturbed J3 with p near z^2; perturbed J2 (+) [c] with p near z; + near-normal A with p-roots engineered between eigenvalues; companion-like + strongly coupled triangular A with lacunary p = z^3 + noise; elliptic-W + configurations with p-roots pushed just outside the boundary at spread + angles (multiple near-coincident |p| peaks); plus a real slice and a + random baseline. +3. **Optimizer.** Derivative-free (Nelder–Mead restarts + compass polish on a + 20-real-parameter Schur-form chart) instead of Chebfun + BFGS / + gradient-sampling. Different method, different stall behavior — an + endpoint here is only called a candidate local maximum after an explicit + *escape probe* (240 random perturbations at 4 scales + one fresh + Nelder–Mead restart) fails to ascend. +4. **Certified endpoints.** Every census endpoint, rounded entrywise to + denominator 256, gets a rational-arithmetic certified enclosure of its + ratio from the tier-0 harness (`ratio.py`), and the top endpoints are + re-verified by the independent route (`verify_ratio.py`, + Faddeev–LeVerrier + Sturm). GO's computations are floating-point + throughout. + +Parametrization note: the ratio is invariant under unitary similarity, so A +is kept upper triangular (Schur form, WLOG) — 6 complex entries; p has 4 +complex coefficients, normalized to unit norm inside the objective (pure +gauge). 20 real parameters total. + +## What was done + +All commands run from the blind working copy root. + +1. **Harness self-tests** (both pass): + + ``` + python3 harness/crouzeix/ratio.py --selftest # 1.8 s + python3 harness/crouzeix/verify_ratio.py --selftest # 2.6 s + ``` + +2. **Cost measurement before sizing** (`explore/measure_cost.py`): one + certified enclosure of a dense 3x3, denominator-32 configuration with a + degree-3 polynomial costs **0.35 s** (tol 1e-6, 16 half-planes) to + **0.82 s** (tol 1e-9, 32 half-planes); denominator-256 endpoints cost + about 1 s. So certifying every endpoint of a few-hundred-point census is + affordable; the float exploration dominates the budget instead + (~19 s per optimization start at ~4000 objective evaluations, three + workers running concurrently beside two sibling agents). + Census sized accordingly: target ~300 starts. RECORDED NARROWING: the + original design wanted ~600+ starts; concurrency made the per-start cost + ~2x the pilot estimate, so the interleaved run was cut at ~240 records + (balanced across families by interleaving) plus 39 records from a first + truncated launch (family-ordered, so family-biased; kept, labeled). + +3. **Float validation.** The float ratio evaluator reproduces the literature + values at this slice: (J3, z) -> 1.41421 (sqrt 2), (J3, z^2) -> 2.0000, + (J2 (+) [0.2], z) -> 2.0000, normal matrix -> 1.0. + +4. **Known-structure anchors, certified by both routes** (calibration for + what "certified near 2" can look like at 80 half-planes): + + ``` + (J3, p=z^2): ratio in [1.980557335, 2.25], both routes AGREE + (J2 (+) [1/4], p=z): ratio in [1.990266652, 2.0], both routes AGREE + ``` + +5. **Census.** + + ``` + python3 explore/census.py --run 25 --seed 4 --out data/census_i4.jsonl # x3 workers, seeds 4,5,6 + # earlier truncated launch: --run 30 --seed {1,2,3} -> data/census_w{1,2,3}.jsonl + ``` + + 286 starts total (241 interleaved + 45 from the truncated first launch; + per-family: random 81, jordan2 34, jordan3 36, nearnormal 34, companion + 34, peaks 34, real 33 — random is overweighted by the truncated launch). + Each start = NM(step .2) -> NM(step .05) -> compass polish at 192 + boundary samples -> diagnostics at 512 boundary samples. + +6. **Classification + escape probe.** + + ``` + python3 explore/classify.py data/census_*.jsonl # -> data/classified.jsonl + ``` + +7. **Certification of every endpoint** (den 256, tol 1e-6, 32 half-planes) + and independent verification of the top endpoints: + + ``` + python3 explore/certify_endpoints.py data/classified.jsonl --den 256 \ + --out data/certified.jsonl --top-verify 8 + ``` + +8. **Structural detail of probe survivors.** + + ``` + python3 explore/survivors_detail.py data/classified.jsonl below2-max + ``` + +## Outcome + +**Status: EVIDENCE** — scoped to dimension 3, polynomial degree <= 3, and +the search design above (286 starts across the 7 named families, escape +probing as the local-maximality test). Nothing here is a claim about the +conjecture. + +**Census verdict: every surviving local maximum sits at already-known +extremal structure. Nothing new appeared.** + +Endpoint classes (286 starts, after escape probing; full data in +`data/classified.jsonl`, per-endpoint certified enclosures in +`data/certified.jsonl`): + +| class | n | meaning | +|---|---|---| +| stall | 175 (61%) | escape probe found ascent: an optimizer artifact, not a local maximum | +| near-2 (double eig) | 48 | ratio > 1.97, two eigenvalues coalescing: J2 (+) [c]-type basin | +| near-2 (other) | 44 | ratio > 1.97, eigenvalues still separated, W(A) near-disc (median disc residual 0.03) | +| near-2 (triple eig) | 2 | ratio > 1.97, all eigenvalues coalescing: J3-type basin | +| ratio exactly 1 | 13 | genuine local maximum at value 1 | +| strictly between 1 and 1.97 | 4 -> 0 | all four fell to deep probing (see below) | + +- **The 13 ratio-1 endpoints** all show one eigenvalue of A lying *on* the + boundary of W(A) with the single |p|-peak at it (boundary margins < 1e-3 + of diam W in `explore/survivors_detail.py` output). This is the + ice-cream-cone configuration Greenbaum-Overton describe for their + locally-minimal value 1 — a **rediscovery, with citation** (Calcolo 2021). +- **The 97 near-2 endpoints** have near-circular W(A) (median disc residual + 0.002-0.03 across the three subclasses) and climb toward the known + Crouzeix pairs; the eigenvalue-coalescence classes match (J2 (+) [c], z) + and (J3, z^2). The "other" subclass has not yet collapsed its eigenvalues + — consistent with the extremal manifold being approached along + non-defective matrices, with coalescence only in the limit. +- **The 4 intermediate candidates** (float ratios 1.96764, 1.96026, 1.95884, + 1.78117) all **escaped under deep probing** (`explore/deep_probe.py`: + 512-sample objective, 1200 random perturbations at 5 scales, 3 jittered + NM restarts, fine compass) with ascents of 0.005-0.006. Verdict: stalls. +- **Certified enclosures: all 286 endpoints** (rounded to denominator 256) + were certified by `ratio.py` (tol 1e-6, 32 half-planes, ~0.75 s/point, + 214 s total): **0 refutations**; the largest certified lower bound on any + endpoint ratio is **1.969725** (compare the same pipeline's certified + lower bound of 1.974 at the exact 2x2 Jordan pair — i.e. the near-2 + endpoints are as close to 2 as the 32-gon slack allows one to certify). + The **top 8 endpoints by float ratio were re-verified by the independent + route** (`verify_ratio.py`: Faddeev-LeVerrier + Sturm): all 8 AGREE, and + both routes agree the refutation test fails. +- **Not seen:** the published intermediate locally-maximal values at this + slice — reciprocals 0.698 and 0.844, i.e. ratios ~1.433 and ~1.185 + (values `[L]`-transcribed from a machine summary of arXiv:2105.14176; not + read from the paper itself) — never appeared as endpoints of any of the + 286 basins. No start family here reproduced them. + +## Why it failed / what survived + +This was a census, and the dead-end is the finding: + +- **All basins from this design terminate at known structure** (ratio-2 + Jordan-type pairs and ratio-1 ice-cream-cone points). No new extremal + structure at n = 3, deg <= 3 from these seven families. +- **The design did not reach the published intermediate maxima.** GO's + intermediate values came from ~half a million heavy-tailed random starts; + 286 structured starts found none of those basins. So this census can say + nothing about them beyond: their basins, if genuinely locally maximal, + are small relative to this start distribution. That is the main + limitation of the design, and it was forced by the measured cost + (recorded narrowing in "What was done", item 2). +- **What survived methodologically: the stall rate.** 61% of derivative-free + maximization runs terminate at points that *look* like interior local + maxima (ratios spread over [1.0, 1.97], certified lower bounds up to + 1.94) but escape under deeper probing. A census that skipped the escape + probe would have reported dozens of fake "new local maxima". Any future + landscape claim on this problem should be required to carry an + escape-probe pass; the 4 deep-probed examples here are the concrete + cautionary cases. +- SPECULATION: at n = 3, deg <= 3, the only locally maximal values of the + Crouzeix ratio reachable with non-negligible basin measure from generic + or J-structured starts are exactly 1 and 2. (The published intermediate + values would then be either heavy-tail-accessible only, or nonsmooth + stationary values that a stronger escape test might also dissolve — + untested here.) + +## Leads generated + +1. **Attack the published intermediate maxima directly** (falsifiable): + build starts matching GO's description of their intermediate minimizers + — eigenvalues well separated, the three roots of p nearly coincident, + several active |p|-peaks — and run the deep-probe test on the resulting + endpoints. If they escape, there is a discrepancy with the published + local-minimality claim worth escalating (first re-checking the paper + itself, not the `[L]` transcription; informed-side task). +2. **Tighter certification of a single near-2 endpoint**: den 1024, + dir_bound 7+, tol 1e-12 should certify an endpoint ratio inside + [1.99, 2.001]; cost measured ~5-10 s/point, entirely affordable. +3. **Ridge geometry of the stalls**: the 175 stalls concentrate where the + active set of max_{dW}|p| changes cardinality (3-4 near-coincident + peaks). Characterizing that ridge could explain which nonsmooth + stationary values exist between 1 and 2 — and whether GO's intermediate + values live on it. +4. **Harness lead**: an escape-probe utility (random multi-scale + fresh- + simplex restart) is generic for any optimization-landscape claim in this + repo and could move into `harness/` as tier-0 tooling. + +## References + +- M. Crouzeix, C. Palencia, *The numerical range is a (1+sqrt 2)-spectral + set*, SIAM J. Matrix Anal. Appl. 38 (2017). +- A. Greenbaum, M. L. Overton, *Numerical investigation of Crouzeix's + conjecture*, Linear Algebra Appl. (2018); preprint + optimization-online 5703 (2016). +- A. Greenbaum, M. L. Overton, *Local minimizers of the Crouzeix ratio: a + nonsmooth optimization case study*, Calcolo 58 (2021); arXiv:2105.14176. + Key published finding this census must be compared against: minimizing the + reciprocal ratio ||p||_W / ||p(A)|| from heavy-tailed random starts, they + found (i) 1 is a frequently occurring locally minimal value, (ii) many + locally minimal values strictly between 0.5 and 1 exist, often the same + over real and over complex parameters, and (iii) the smallest locally + minimal value found is always 0.5 — i.e. in this record's convention, + local maxima of the ratio at 1, at many values strictly between 1 and 2, + and at 2, never above. +- Tier-0 harness: `harness/crouzeix/ratio.py`, `harness/crouzeix/verify_ratio.py`. diff --git a/problems/crouzeix/data/census_i4.jsonl b/problems/crouzeix/data/census_i4.jsonl new file mode 100644 index 0000000..8d5bcda --- /dev/null +++ b/problems/crouzeix/data/census_i4.jsonl @@ -0,0 +1,79 @@ +{"family": "random", "seed": 400000, "x": [1.1637546894699131, 1.3387571672177243, 1.6092230250910644, -0.6324386977805513, -0.5786510560275908, 1.3807163847105475, -1.2770144373189805, 0.3327930226284563, 0.6855809351172977, 0.6567599551279006, 0.8653677728707757, 1.6492690818808908, -0.4680017551301743, 1.9717084498434407, 1.0749568514281624, -1.1195807394928647, -1.1194896571507935, 0.541507939218701, 1.5546265891994846, -0.9748372026574621], "ratio96": 1.4733466106055264, "ratio_final": 1.474585189086896, "diag": {"ratio": 1.474585189086896, "nonnormal": 0.6815660377178587, "eig_gaps": [0.45026467441734236, 0.5220817275692271, 0.7680839058690033], "discness": 0.23395279665038896, "peaks": 3, "deg": 3, "diamW": 3.8708585980884043}} +{"family": "jordan3", "seed": 401000, "x": [0.16813274325574726, 0.1267920593445052, -0.2269579418812387, 0.29234757829988345, -0.45320988643118, 0.2065849867685672, 1.9578803040341368, -0.38386813569009537, -0.42811876624542855, 0.026407176195512028, 2.14531753556515, -0.039140190899580526, -0.12054782373550071, -0.21518652393552956, 0.1798048834730397, -1.617509404769688, 2.5115610147398932, -1.4266432013885009, 0.29729074082271817, -0.12175554302440633], "ratio96": 1.973242533081747, "ratio_final": 1.974033911016823, "diag": {"ratio": 1.974033911016823, "nonnormal": 0.7490028296412479, "eig_gaps": [0.08074737656884254, 0.14295757990216038, 0.2090576114453739], "discness": 0.005388684247591025, "peaks": 3, "deg": 3, "diamW": 2.9965194024681}} +{"family": "jordan2", "seed": 402000, "x": [0.2118205430836373, -0.12011191791608411, -0.1771870327824911, 0.02853799878724763, 0.1953939747827479, 0.34748575426978234, 1.4142097830310956, -0.29499659515509274, 0.33319460297083336, 0.36328377180905314, -0.028813029677675572, -0.0007975654446331134, -0.08344206619356137, -0.02209705102198633, 1.739516935276686, -0.49861528316727655, -0.020244626630011864, -0.14025816300462812, -0.03301056786922761, 0.01001330726583325], "ratio96": 1.9724002360395447, "ratio_final": 1.972475733235554, "diag": {"ratio": 1.972475733235554, "nonnormal": 1.3255588791027426, "eig_gaps": [0.2630065212921424, 0.29549658029597026, 0.30974893270759923], "discness": 0.006665045235292756, "peaks": 3, "deg": 3, "diamW": 1.583389266309631}} +{"family": "nearnormal", "seed": 403000, "x": [-0.8340652029109421, 1.3776805332489246, -0.89902307819048, 0.003539677027255185, -0.14904558094480436, 1.1048924801555073, -0.9424538495244978, -0.31354668908244854, -1.4839312764248733, 0.18165676074196407, 0.14732872225860344, 0.9197300442040808, 2.945779862567481, 1.4171345279649459, 2.8282302783765125, -0.8648540507100315, 3.6519454847211708, 0.7840822699605925, -0.06953594025205653, 0.5693541951192229], "ratio96": 1.636683224465937, "ratio_final": 1.6399400334716803, "diag": {"ratio": 1.6399400334716803, "nonnormal": 0.5800868283122488, "eig_gaps": [0.2974564409894917, 0.5375404212731334, 0.5549751764635916], "discness": 76248354153.57133, "peaks": 3, "deg": 3, "diamW": 2.4788051611325033}} +{"family": "companion", "seed": 404000, "x": [-0.302070942800376, 0.7186957327412735, -1.220431966355091, -0.28302913538871954, 1.1922976298060095, -0.3861345695243055, 0.006742010288531117, 0.9868143418417582, -2.9363934560295832, -0.8461785373454943, 1.4725339643484996, -0.8117568349307664, 0.24565062629463483, -0.19163772035101503, 0.3483504550341804, 1.8304935890793452, 0.10378057082829252, -0.018477927634751905, -0.0024398583052655607, -0.032012284175236785], "ratio96": 1.9233604487589435, "ratio_final": 1.925728641730908, "diag": {"ratio": 1.925728641730908, "nonnormal": 1.3117474161398588, "eig_gaps": [0.31479708115845295, 0.4304908255783476, 0.5593980397743075], "discness": 0.047766850371888075, "peaks": 4, "deg": 3, "diamW": 4.317018420032665}} +{"family": "peaks", "seed": 405000, "x": [-0.9829424448880801, -0.7162593769905634, -0.24827329725911917, 0.10670932013826717, 0.8170578187790367, -0.7850402123191147, 1.6134320656710042, 1.0544961635406118, 0.18325311319467358, -0.10721074717367185, 1.7917002484353386, -0.07948521792676548, -0.6293219268544203, -1.6296960493545187, -1.8190271942750518, 0.6733837339392079, 0.0859049105321592, 1.6120233534071011, -0.4721426411982612, 0.14468568082262584], "ratio96": 1.9591307275493353, "ratio_final": 1.9590367342496156, "diag": {"ratio": 1.9590367342496156, "nonnormal": 0.7301317237937558, "eig_gaps": [0.343845566856024, 0.4330225430648908, 0.5614416708015036], "discness": 0.07597811246029411, "peaks": 5, "deg": 3, "diamW": 3.2083722848565004}} +{"family": "real", "seed": 406000, "x": [1.5230879030064817, 0.1917667455752054, 1.1691083250469634, 0.040517370157498095, 1.2261733719182404, 0.03094809369655139, -0.0578445295265749, 0.30359123280601236, -0.7908814716316774, -0.30150336530603894, -0.14269043710767776, 0.008289110011652815, -1.5339787728821053, 0.10868625788913289, -0.6099159703026269, 2.892241883098684e-05, 1.6065284472280585, -0.49408933258631627, -0.2823879295699706, 0.23888531386791997], "ratio96": 1.9504599929031357, "ratio_final": 1.9563442656411278, "diag": {"ratio": 1.9563442656411278, "nonnormal": 0.20410747695370537, "eig_gaps": [0.059857837384671014, 0.3493181045882263, 0.3982177786335099], "discness": 9790131382.9728, "peaks": 3, "deg": 3, "diamW": 0.9666540940055174}} +{"family": "random", "seed": 400001, "x": [1.2534022026553093, -0.7839057835795689, -0.6320330927185923, 1.163510247712968, 0.4654765054130657, -0.7511230684899993, 0.7289954445143618, 1.9911578814058042, -1.871329088365571, 0.4283283932361738, 2.200024191893569, 0.6452098054736204, -1.115536301921391, -4.782942911108577, 2.771124032085873, -0.8636181430322365, -4.048686985588349, -0.1899531405733032, -0.21753776402531297, 0.44097331132699996], "ratio96": 1.9495687744872776, "ratio_final": 1.9495289236168243, "diag": {"ratio": 1.9495289236168243, "nonnormal": 0.741336075881184, "eig_gaps": [0.16960800350709668, 0.4746412734077458, 0.5829740901127468], "discness": 0.10784089552554417, "peaks": 3, "deg": 3, "diamW": 4.649588289840303}} +{"family": "jordan3", "seed": 401001, "x": [0.459123380682876, 0.26786964995147866, 0.11201371160172235, 0.0886819681920511, -0.4774684099526276, 0.3787335545823378, 1.251189816791397, -0.4034200451877263, -0.03984505200977382, 0.11594737599622738, 1.1883934669823124, 0.22274549982880473, -0.31537636331976127, 0.04990318284169133, -0.02939474074717155, -0.7597996794741645, 1.316283271329167, 0.11335949855014238, -0.019377717460602698, -0.31660698732469605], "ratio96": 1.9895992576924333, "ratio_final": 1.989554662554931, "diag": {"ratio": 1.989554662554931, "nonnormal": 0.7034575313536575, "eig_gaps": [0.19239505283938285, 0.3235760914393243, 0.46451297853252504], "discness": 0.055174570508538585, "peaks": 4, "deg": 3, "diamW": 2.0303639786076593}} +{"family": "jordan2", "seed": 402001, "x": [-0.7025273628037407, -0.4397797168352926, 0.2532139695733695, 0.01625216637404006, 0.290104253732692, -0.16539956349199142, 1.6977196047063063, -0.5864427883957316, 0.22484896447734037, -0.8462954944338525, 0.058045183643668896, -0.0881592432452249, 0.28337203601165384, 0.023275679841948796, 0.6935824464036187, -0.6038792672100404, -0.05588925138731571, 0.031973418086524555, -0.004829612918725079, 0.051584593890371844], "ratio96": 1.9694414648093004, "ratio_final": 1.969392851389826, "diag": {"ratio": 1.969392851389826, "nonnormal": 1.3207392998078549, "eig_gaps": [0.08142208947786002, 0.4523795695427998, 0.4651663729373806], "discness": 0.061776182893447475, "peaks": 4, "deg": 3, "diamW": 2.276529260841256}} +{"family": "nearnormal", "seed": 403001, "x": [-0.009475797701735617, 0.3974203276399666, -0.4940486298404265, 1.024892955655842, 0.12297324047548269, 1.6261423208733339, 0.3998123904387274, 0.3831409906146256, -1.353447570891378, -0.2885291068378949, 0.5582969342576793, -0.5075173362106897, 6.082011666918385, 6.989655394010462, -9.945138419198493, 5.250400975088873, 1.0624609868592616, -4.551583688212414, 0.838010192856808, -0.08642697386836584], "ratio96": 1.6147835185003712, "ratio_final": 1.667019965562423, "diag": {"ratio": 1.667019965562423, "nonnormal": 0.7115820963332398, "eig_gaps": [0.3820243680677492, 0.4151380147068706, 0.5955104162876467], "discness": 52798461394.470085, "peaks": 2, "deg": 3, "diamW": 2.075261754638286}} +{"family": "companion", "seed": 404001, "x": [0.559644580272846, 1.2092786314653692, 0.34640786001761215, -1.3134996879945944, -0.01278807055400431, -1.0386209380715925, 2.2028205130881546, -2.4049062555848604, 0.4362603005498601, -1.6093874924190836, 1.6876497670606865, 1.9354975281553213, 0.29433088194514895, -0.29237929129913853, 0.019727497321109115, 1.0135081101970773, 0.7666431539448728, -0.619347057000355, 0.007807150569793326, 0.12466181742766147], "ratio96": 1.8610581695474009, "ratio_final": 1.9249874758765917, "diag": {"ratio": 1.9249874758765917, "nonnormal": 0.8085260109936573, "eig_gaps": [0.0878073052446734, 0.45031838777416466, 0.49150051468002887], "discness": 0.05929225666067442, "peaks": 3, "deg": 3, "diamW": 5.151111900510527}} +{"family": "peaks", "seed": 405001, "x": [-0.02173524795549614, -0.39601778332558846, 0.18692974993208444, 0.5299692118620043, 0.4585859149183178, 0.02576874664711823, 0.988407888166897, 2.3552294155351343, -0.5548415272229308, -0.1488470363138631, 2.5064878763822014, -0.3423311152537711, -1.3123293255189705, 0.2653458130092346, 3.293248397562227, 0.7199416854723517, -7.314313228557157, 0.7841829763824721, -0.5179216052564788, 0.05726850429804414], "ratio96": 1.988013301339618, "ratio_final": 1.987997920438502, "diag": {"ratio": 1.987997920438502, "nonnormal": 0.727576425892596, "eig_gaps": [0.15222560481144842, 0.16990120226686875, 0.2522908173898413], "discness": 0.011653711960387395, "peaks": 3, "deg": 3, "diamW": 3.7623506029783775}} +{"family": "real", "seed": 406001, "x": [-0.988113351149956, -0.004215695963268935, -0.10783234770594591, -0.014712474627010544, -0.74256409378083, -0.05630888044065854, -1.2366422453867592, -0.1867344033911941, -0.2210236323594455, 0.02577977929101017, -1.3427010258243375, 0.10384361785827825, 0.31876850562860204, 0.027435905379162365, 1.059872620387306, 0.054812150354345585, 0.5836337044098387, -0.032548936295118014, -0.22242323129668862, -0.03748991310826398], "ratio96": 1.987049615233899, "ratio_final": 1.987039203566344, "diag": {"ratio": 1.987039203566344, "nonnormal": 0.5737179953902536, "eig_gaps": [0.12258409862794858, 0.31063945689876693, 0.42992036733957695], "discness": 0.08006903626634833, "peaks": 3, "deg": 3, "diamW": 2.04768987870431}} +{"family": "random", "seed": 400002, "x": [2.2965814774339073, 3.2318532328058307, 1.481088281830253, -2.7417876069041736, -1.9397073521240773, -0.06358556137594344, 1.1101119646631696, -5.307475412800426, 4.506333248459718, 0.7605650059422844, -4.396211518723932, -0.6559125604562352, 3.1606539288619127, -0.6530344188555036, 1.3958658774037316, 1.8692336401755858, -1.7928349216148893, -2.8902811826895602, 0.21989494617164101, -0.005988294238791124], "ratio96": 1.8845034327679335, "ratio_final": 1.8855779399806656, "diag": {"ratio": 1.8855779399806656, "nonnormal": 0.8095592008327809, "eig_gaps": [0.429862475460481, 0.5310464177097892, 0.5965395757537864], "discness": 0.04318252542616364, "peaks": 4, "deg": 3, "diamW": 10.106701664053938}} +{"family": "jordan3", "seed": 401002, "x": [0.20139600696814852, 1.064168733153285, -0.35321485651948403, -0.34082059765571016, -0.28413044129278153, -1.0680420728159072, 0.33249151280454625, -0.026997359810870012, 0.3338861048819237, 0.045460280544681014, 1.1809106299276404, -0.09973370120148352, 0.589510188668531, -0.0727329591203819, 0.10951727492730978, -0.1279616031082924, 0.8368803277298831, 0.4908092341074186, 0.16733697682434046, -0.24588480582936134], "ratio96": 1.7687693957512514, "ratio_final": 1.7552977507389464, "diag": {"ratio": 1.7552977507389464, "nonnormal": 0.6315317711148664, "eig_gaps": [0.28391052032307385, 0.587060083806575, 0.8499069423502256], "discness": 0.32449348442431614, "peaks": 4, "deg": 3, "diamW": 2.5729780520485184}} +{"family": "jordan2", "seed": 402002, "x": [-0.17361558324087356, 0.13403392790124413, -0.016709072423486196, -0.215016720707923, -0.1611948185527029, 0.3040663167515961, 1.2358625671654415, -0.059345765614480156, -0.5097572326442035, -0.022739466242883907, -0.11796971001040496, 0.24899904588950472, 0.14305019883988235, 0.10641089683623506, 1.7472768250891861, 0.19197298585687284, 0.01907917764283925, 0.02010315217455351, 0.04530198215927966, -0.03748807907316161], "ratio96": 1.9953873588599265, "ratio_final": 1.995382827290676, "diag": {"ratio": 1.995382827290676, "nonnormal": 1.3096591973131966, "eig_gaps": [0.12080187113680109, 0.27116893648586154, 0.3817924270528878], "discness": 0.013093141115402059, "peaks": 3, "deg": 3, "diamW": 1.4112815337368492}} +{"family": "nearnormal", "seed": 403002, "x": [1.098821609310131, -0.9408546793452492, 0.8975718933055052, -0.9767392254631577, 1.1196559334782767, -1.2100051294064582, 0.09724297846315663, 0.05772732859267432, -0.09548194490342883, -0.2589530578067273, 0.697017583581543, 0.04679472355237651, -0.4097027358661447, 1.2392651365918743, -0.9641392252067646, -4.6365414160419345, -2.376110461528448, 3.988833144900621, 1.1138723542647768, -0.0040385892590619425], "ratio96": 1.8933692608818016, "ratio_final": 1.8941728389597947, "diag": {"ratio": 1.8941728389597947, "nonnormal": 0.12372249642697516, "eig_gaps": [0.24712456054766652, 0.32634466071097823, 0.38935463006481746], "discness": 10553153873.514143, "peaks": 3, "deg": 3, "diamW": 0.8272101463846251}} +{"family": "companion", "seed": 404002, "x": [0.16594654557402855, 0.1325588312465653, 0.08509788150556089, -0.1737939455064732, 1.9838025032853912, 1.3296869962587827, -3.450607048506513, 1.7169113393229505, -0.48910795872714563, 0.07336391694463582, 3.0577482792707453, 1.2164242578411335, -0.33213021260968756, 0.09341151192922283, 2.3477713166480587, -5.481605807864188, 0.14966305432022564, 2.775477680130379, 0.1908739794371842, 0.07287145472301515], "ratio96": 1.9711905081361913, "ratio_final": 1.9756485744848153, "diag": {"ratio": 1.9756485744848153, "nonnormal": 0.6756619406061208, "eig_gaps": [0.05524106542609626, 0.3794936520502056, 0.4222540104997716], "discness": 0.05927475397182865, "peaks": 3, "deg": 3, "diamW": 5.735615133008684}} +{"family": "peaks", "seed": 405002, "x": [-1.2357904607336154, -0.02652099922100296, 0.47381033338597156, 0.7246994985785067, 0.7320243321857363, 1.2767440073004872, 1.1724237115692469, -0.1264005301237222, -0.1372222266740568, -0.033456219283891, 1.2764445365839505, 1.4305699752382663, -3.0001061066387638, -0.7337219395312634, -1.5029442367895651, -3.2097927065579706, 2.7760880927291636, -1.6752083917275167, -0.22999947522037434, 0.40491075354533634], "ratio96": 1.8883416000820648, "ratio_final": 1.8892524871320382, "diag": {"ratio": 1.8892524871320382, "nonnormal": 0.5933349916317641, "eig_gaps": [0.1819273360376797, 0.5574308777553952, 0.7045623749860311], "discness": 0.27539248149602147, "peaks": 4, "deg": 3, "diamW": 3.3499562172964015}} +{"family": "real", "seed": 406002, "x": [2.57330113920448, 1.0470074709203887, -3.561689272740095, 1.340685173213159, 2.705031382423333, 1.623867511292111, 2.906300750450799, 1.9280797358702975, -5.782088311032604, -0.6494012370289151, -3.650467216799422, -1.0218439336679426, 4.307155666179382, -3.950286431639208, 2.2948510668011064, 2.7714256911911246, 2.4767759274068704, 0.14123535095861245, -0.16064770571269726, 0.08360002206249093], "ratio96": 1.5307624397625792, "ratio_final": 1.5424697871862874, "diag": {"ratio": 1.5424697871862874, "nonnormal": 0.8027542317247043, "eig_gaps": [0.05630620571677072, 0.584464964993525, 0.5969402622186633], "discness": 0.227202665962716, "peaks": 3, "deg": 3, "diamW": 10.508782956814636}} +{"family": "random", "seed": 400003, "x": [-4.143666743522802, -2.4372769532346457, 2.599104236417321, -2.5585472364811226, 3.0784887547565933, 2.9583409434589676, 2.0384036611578313, -0.38927419147899145, -0.9000404716592671, -2.8228240095436874, 1.7280353191665332, 1.0218940294326686, -1.734139221990316, 2.1333807003688396, 0.6053124842639921, -3.658212048668629, -0.6579889143987272, -0.8791447075668719, 0.12500420219225294, -0.1637261875218362], "ratio96": 1.3338882421899658, "ratio_final": 1.3693221045958268, "diag": {"ratio": 1.3693221045958268, "nonnormal": 0.6444518263110308, "eig_gaps": [0.5577616311774807, 0.6792500292747193, 0.9080130622086073], "discness": 0.19113494168280235, "peaks": 3, "deg": 3, "diamW": 9.928393283842611}} +{"family": "jordan3", "seed": 401003, "x": [-0.3238977360555537, -0.015273286900158209, 0.7679847592702946, 0.39741178137707, -0.33366993800244116, -0.42840451207797126, 1.7695251654058153, 0.359136056484542, -0.5196049561163365, 0.0731330642687805, 1.6202210158073578, 0.7468723872367691, -0.023356757887338632, -0.44829361607431006, 0.07156672719085405, 0.2524214058472434, 2.1772446915116728, 0.47562544413736463, -0.310598264286569, -0.0082592132432188], "ratio96": 1.9644149141611291, "ratio_final": 1.9646022916763135, "diag": {"ratio": 1.9646022916763135, "nonnormal": 0.725743885901117, "eig_gaps": [0.1392445027340264, 0.39331402806004195, 0.4639206613911675], "discness": 0.049055151561915825, "peaks": 3, "deg": 3, "diamW": 2.967778093112101}} +{"family": "jordan2", "seed": 402003, "x": [0.29820223367825005, -0.008916112291766053, -0.11147867905317213, -0.025942098814447076, 0.3211006336340839, -0.0008192199266728213, 0.8464313548479238, 0.3946673495363116, 0.6786352120706169, 0.23592157168362995, -0.37342309988386885, 0.11048073457228026, -0.1509486910263062, -0.026962800381941588, 1.9153872639476635, 0.3789983778752231, 0.004024150203612121, 0.015252897226587066, -0.11790192348897544, 0.018846940685068283], "ratio96": 1.9942133187180406, "ratio_final": 1.9942564539334426, "diag": {"ratio": 1.9942564539334426, "nonnormal": 1.3012771164566952, "eig_gaps": [0.01866508222205502, 0.31511022429902275, 0.3329959685578438], "discness": 0.02018950040083996, "peaks": 4, "deg": 3, "diamW": 1.3012416645387301}} +{"family": "nearnormal", "seed": 403003, "x": [0.612775113115535, -0.08456720812126954, 0.410604527368326, -0.7011073493200497, 1.1737415067311219, 1.2141837847068544, 0.5985521648646562, 0.9586704491494213, -0.051205778465194474, -0.9226335848585832, 1.0247151843479783, -0.1991496220799942, 0.9757529384266178, -1.0640288768078818, -1.3833800631261606, 0.5269437524110984, 0.0633362229041409, -1.0569570053578814, 0.7032585665808946, 0.026458805866413693], "ratio96": 1.6710239962764462, "ratio_final": 1.7126386013132484, "diag": {"ratio": 1.7126386013132484, "nonnormal": 0.663967925161178, "eig_gaps": [0.2350664768114006, 0.5125348464667521, 0.746936262867808], "discness": 0.4107229355255806, "peaks": 3, "deg": 3, "diamW": 2.760244871668558}} +{"family": "companion", "seed": 404003, "x": [-0.6191856827551976, 1.3885205312751767, -0.05488967841986589, 1.2534901237661673, -0.48665806964423997, -0.7145381435425232, 1.0574906314230224, -0.232345917608567, 0.45362983778029375, 0.6297498935215046, 1.3343000711559556, -0.15531167290919334, -0.012031117033858826, 0.14395729658669987, 0.054564335667510594, 0.42637899247333666, -0.06807910656527888, -0.01642576497166228, 0.18787057970143517, 0.09863440923388167], "ratio96": 1.6762682993505984, "ratio_final": 1.6764003782028387, "diag": {"ratio": 1.6764003782028387, "nonnormal": 0.6337571682462164, "eig_gaps": [0.19819659032163844, 0.6882366572643926, 0.7197975339700344], "discness": 0.2508189465341587, "peaks": 4, "deg": 3, "diamW": 2.927531921149381}} +{"family": "peaks", "seed": 405003, "x": [0.06488670547085905, 0.6571318627495162, -0.0719574579837327, 0.6868121035437811, 0.23314002127388653, 0.9519772817956196, 0.031624631256464336, 0.011265243388016448, -0.24202815946953418, -0.14634732226761502, 0.9979310789869189, -0.3292474460851618, -0.26075654194429626, 2.310178598556929, -2.4132684724596905, -0.0008316773648728481, -0.9222523682132648, 0.013481874374360785, 0.12310139346277692, -0.18251964856455416], "ratio96": 1.9708704792724532, "ratio_final": 1.9712314269731561, "diag": {"ratio": 1.9712314269731561, "nonnormal": 0.5841417624508887, "eig_gaps": [0.1206200175609261, 0.2924276350145401, 0.348203580562967], "discness": 12760030514.373123, "peaks": 3, "deg": 3, "diamW": 1.1608841043969915}} +{"family": "real", "seed": 406003, "x": [-0.36948292612956335, -0.1286238434820614, -0.24586482803610799, -0.08388095442261244, -0.26245537420859466, -0.023813641866055585, -0.03414374399308741, 0.22665662039408874, -0.6715005291132485, 0.270489651814715, 0.4013567471189161, 0.23392967975543555, -0.5451519702980452, -0.032033646760559684, -1.7609553833314866, 0.4621121664432868, 0.20409457512107254, -0.07943699514189187, -0.08347471948523372, -0.1421523762464439], "ratio96": 1.9821370321564569, "ratio_final": 1.9829369879028436, "diag": {"ratio": 1.9829369879028436, "nonnormal": 1.03491657023867, "eig_gaps": [0.06900230132034489, 0.14557125140392385, 0.16587222792594403], "discness": 0.021745888893262677, "peaks": 3, "deg": 3, "diamW": 0.9031054774697571}} +{"family": "random", "seed": 400004, "x": [-1.0238424224522593, -0.19073810199422947, 0.38405792692029905, -0.8022105541133202, -1.1824458412509555, -1.4664708502019321, -0.38013318900792903, 1.5993998908247278, 0.41810270410820666, 0.7244130742447255, 0.06938873745905763, 1.1685220515834667, 2.214099374005669, 1.2490767833295962, 2.97570522561518, -1.8890249076990755, 1.1799754611713338, 0.3781940080305125, 1.3684438531240062, 0.36489145162399605], "ratio96": 1.6736653118177551, "ratio_final": 1.6791706927654377, "diag": {"ratio": 1.6791706927654377, "nonnormal": 0.5741465651442422, "eig_gaps": [0.4641809136863659, 0.5542327068120335, 0.6143762580333431], "discness": 0.15910230762251326, "peaks": 3, "deg": 3, "diamW": 2.769510686511074}} +{"family": "jordan3", "seed": 401004, "x": [-0.04353481431211606, 0.10555704414719758, 0.2568938143382402, 0.3658538585599047, -0.25896148447226996, -0.17881485696830512, 0.8741741170680681, -0.3553568960972869, -0.34210399339251807, 0.3930810307242405, 0.8677819627013804, 0.35981953697070346, -0.02732395088410726, -0.07481517742477534, -0.036059169539717426, -0.3521367747310453, 1.3646254433352318, -0.07179681221346128, 0.20009398981808957, -0.2933203248633396], "ratio96": 1.9913094441734593, "ratio_final": 1.9912826371971442, "diag": {"ratio": 1.9912826371971442, "nonnormal": 0.7199149575647555, "eig_gaps": [0.22182487078105834, 0.24716198801164574, 0.466447391374071], "discness": 0.04583853696790835, "peaks": 4, "deg": 3, "diamW": 1.608285217490507}} +{"family": "jordan2", "seed": 402004, "x": [-0.06613300018958321, -0.04404276550997127, -0.2665646803887126, -0.47296028699590464, -0.0017192160906624497, 0.18808928986884255, 0.42691792463692907, -1.432704556380961, 0.73189821338082, 0.16160792587049322, 0.2905947714996434, -0.26281152042302935, 0.25038327725847065, 0.08635932795656015, 0.6273659131165013, -0.5273341440183246, 0.05863062251495771, 0.019253481824348717, 0.022355484544824222, -0.003350326751671481], "ratio96": 1.9910819828899966, "ratio_final": 1.99108154963197, "diag": {"ratio": 1.99108154963197, "nonnormal": 1.3116169295423288, "eig_gaps": [0.13484187675173773, 0.2649992378570075, 0.3986038105723817], "discness": 0.022120510597794994, "peaks": 4, "deg": 3, "diamW": 1.7865618705069537}} +{"family": "nearnormal", "seed": 403004, "x": [-0.6411055993450849, -0.5833352267880088, 0.6617136281014167, -0.7976375265777604, 0.06855979038471052, 2.0122028391320224, 0.22852030800136983, 0.09863030345176402, 8.656574995424773e-06, 9.070714957836103e-06, 1.4564311662302722e-06, 1.0320777078853687e-05, 0.16897081073411543, -0.1747762849859956, -0.15270078965358894, -0.21530395805557664, 0.17654620631582502, 0.5469461325493061, 1.006524511384689, 0.1133941856499105], "ratio96": 0.99999999996982, "ratio_final": 0.9999999999820248, "diag": {"ratio": 0.9999999999820248, "nonnormal": 0.07960469144573055, "eig_gaps": [0.4547942128992848, 0.9268639393750735, 0.9891959366116936], "discness": 0.48054407625096046, "peaks": 1, "deg": 3, "diamW": 2.9031306318763153}} +{"family": "companion", "seed": 404004, "x": [-1.374188638444159, -1.4800292498941063, 0.7468317690637432, 1.1795774726458563, 0.5203696347398206, -0.48580480017974326, 3.128797755130753, 3.3302014643679474, -3.1883551684073073, 0.43717988555093124, 4.446566712056727, 0.06526209498533297, 0.4497793503392589, -0.4412636131613182, -1.5833072302981062, -0.6829246160014771, -5.086025595072723, 8.928486442249206, 0.38878193718169407, 0.7143814827551975], "ratio96": 1.9209336267040593, "ratio_final": 1.9219565692814136, "diag": {"ratio": 1.9219565692814136, "nonnormal": 0.8032667532939068, "eig_gaps": [0.2163069879970563, 0.27536451595623446, 0.4378109500516135], "discness": 0.032662213097169795, "peaks": 3, "deg": 3, "diamW": 7.770017767143024}} +{"family": "peaks", "seed": 405004, "x": [-1.1623865848615051, 0.6509958491850696, -0.5003360230842877, -0.020101183726879884, -0.5036340490734614, -0.03659664609401815, 1.295354048316115, 1.185393715540028, 1.2211443077978765, 1.164968203505442, 0.1356042721765338, 0.2792774311092724, -2.8360542513331763, 2.346435049043228, -3.85639244226409, 0.9658545591901251, 0.22344365455954474, 0.21076864046464272, 0.0497471499238199, 0.08357041257100828], "ratio96": 1.9863456648712265, "ratio_final": 1.9863425155864072, "diag": {"ratio": 1.9863425155864072, "nonnormal": 1.0798935920298736, "eig_gaps": [0.006412076312312546, 0.35933221353351663, 0.3629642979288813], "discness": 0.08568097426039863, "peaks": 4, "deg": 3, "diamW": 2.623475915997666}} +{"family": "real", "seed": 406004, "x": [-0.6833724558009195, 0.28590789581862186, 0.8368358877651443, -0.3257315338483946, 0.10432383941304829, -0.5299022869252201, 2.3320326289319167, 0.022248129832376513, -1.3241970803462717, -0.23767502542305508, 2.2000821555742833, 1.0222498764248833, 0.383337091724671, -0.6626213822858326, 0.37451269623021455, -0.715716681206092, -1.9627507185007529, -0.21778252830792333, -0.1822785937833571, 0.1253179709716135], "ratio96": 1.9585816743666855, "ratio_final": 1.9588423767255672, "diag": {"ratio": 1.9588423767255672, "nonnormal": 0.7735703190313216, "eig_gaps": [0.19145450636405284, 0.28551348787183506, 0.4125601991633988], "discness": 0.039589627671257795, "peaks": 3, "deg": 3, "diamW": 3.9718773732626405}} +{"family": "random", "seed": 400005, "x": [-1.480230380410387, -2.7358386026723704, -0.08284297081073352, 0.9112889916367062, 1.1216636038828318, -0.4256150714809218, -2.888519132210435e-05, 3.107237571685124e-06, -1.861481800006378e-06, -7.114232940708941e-06, -0.17769902164069284, 0.3878543746980968, 1.1388411790919544, -0.07150931169589188, 0.10535510326645858, -0.5890000408290148, -0.8217765896373872, -0.7172422981592165, 1.1649345835032374, -0.695842962736767], "ratio96": 0.9999999999664926, "ratio_final": 0.9999999999752113, "diag": {"ratio": 0.9999999999752113, "nonnormal": 0.09195209432355816, "eig_gaps": [0.454513338269815, 0.878853023554804, 0.9864916438477316], "discness": 0.8940725258068817, "peaks": 1, "deg": 3, "diamW": 3.959148141072227}} +{"family": "jordan3", "seed": 401005, "x": [0.08780466826470232, 0.13202629980266084, -0.5378595702294608, -0.1800878564795924, 0.10156906288870116, 0.2933880689784183, 1.2485115814755958, 0.5899745534209925, 0.6888473452569717, 0.1671128388945428, 1.3872332832468959, 0.2411801182270396, 0.04407989151121971, -0.15559033790406324, 0.4600510416361281, -0.1343684200355708, 1.3702500874669106, 0.29762339978443475, -0.22457938476273365, -0.035998214734935474], "ratio96": 1.985795010674685, "ratio_final": 1.9858703539332918, "diag": {"ratio": 1.9858703539332918, "nonnormal": 0.7229324901114336, "eig_gaps": [0.07164501552118119, 0.3093201053822483, 0.3519893193601399], "discness": 0.029618503748748738, "peaks": 3, "deg": 3, "diamW": 2.26041917130981}} +{"family": "jordan2", "seed": 402005, "x": [0.5143643310319261, -0.6196694125444009, 0.4267571563656317, -0.3923463764647759, 0.6713796651131778, -0.8561063513639969, 0.9931120120064981, -0.27910508480680024, 0.0007202774379802669, 0.1049586161966505, -0.05055213341976966, -0.04996901003656072, -0.7063548096244454, 0.6677229579663577, 1.3117775801216105, 0.05423293843785007, 0.05657641294003245, 0.14307545003567024, 0.05418561276463538, -0.04348766628545159], "ratio96": 1.9966997062109073, "ratio_final": 1.996750219923216, "diag": {"ratio": 1.996750219923216, "nonnormal": 0.4824233480878822, "eig_gaps": [0.22836506835912282, 0.2660517104019978, 0.49148976810475037], "discness": 4858400643.76638, "peaks": 3, "deg": 3, "diamW": 1.0668013127831568}} +{"family": "nearnormal", "seed": 403005, "x": [-1.4703097829670948, 1.0390268841869092, -0.7865790963078486, 1.6359837168114044, 1.0877772547716655, 0.7227192705730611, -0.587643242330357, -0.895087453752738, 0.7099360984889834, -0.5174173653380031, 0.874317687508995, 0.022440906111533248, 0.687642722744501, -0.2110050131760136, -1.3888193051476074, 0.23235956995629434, 0.5934988952902791, -2.014221857227639, 1.0375516101990683, -0.018903007991207574], "ratio96": 1.3864290267504016, "ratio_final": 1.4741437723143012, "diag": {"ratio": 1.4741437723143012, "nonnormal": 0.43593527792756276, "eig_gaps": [0.29045294694821994, 0.6672076143691975, 0.8248276902568451], "discness": 1.176211502769235, "peaks": 3, "deg": 3, "diamW": 3.12497823879438}} +{"family": "companion", "seed": 404005, "x": [1.296994532140058, -0.03133698333521135, -1.0254537587940642, -0.6823999212744614, 0.7978940006207635, -1.3030876603584725, 1.8818992824208098, -1.0993878647814024, 0.9364777602181457, -0.8492884361492198, 1.187205650218838, 0.8919213827350465, -0.2314021984774604, 1.125445931538735, 0.770791527432287, 0.4817233328985806, -0.1997482687141089, -0.421763205963558, 0.13744656499692748, 0.06530014923372114], "ratio96": 1.6398986289545667, "ratio_final": 1.6760626219030748, "diag": {"ratio": 1.6760626219030748, "nonnormal": 0.7645869710313646, "eig_gaps": [0.3571611103545714, 0.503540083237188, 0.6305646374399686], "discness": 0.11786194092016526, "peaks": 4, "deg": 3, "diamW": 3.8251119175538046}} +{"family": "peaks", "seed": 405005, "x": [0.7468687901284015, 0.6685977493559463, 0.6814232516318232, 0.08102344001684023, 0.6994839258406955, 0.0036812118537640024, -0.24578171503023244, -0.18377825849778728, 1.4863759641204037, -0.32885719603613306, -0.12256102012716338, 0.06470633806528447, -1.4552996096992192, 1.5948410550879046, 1.0366901638462571, -2.777955189481995, -0.3264345976570303, 0.400913794879848, 0.049967038153144946, -0.15747951694503792], "ratio96": 1.9828310090158723, "ratio_final": 1.9828228939265313, "diag": {"ratio": 1.9828228939265313, "nonnormal": 0.8470431698670341, "eig_gaps": [0.046915587968391605, 0.3492297307349634, 0.3937659771855891], "discness": 7.306914098147017, "peaks": 3, "deg": 3, "diamW": 1.6928908648905165}} +{"family": "real", "seed": 406005, "x": [-0.7281442178773666, -1.7847475897856953, -0.4594757083180713, 0.23761267847591927, -0.21532205165320678, 0.31281523499379704, 1.930518161680395, 0.35606654458631937, -0.036404096858650564, 0.4764740403307944, 2.2671069168100986, -1.0585756535121125, 0.0942644515161769, 0.9806320046735171, -0.07163695250500673, 2.0962117451617033, 0.6442501318745342, 0.7627860529481163, -0.17077023457125762, 0.08271346537950511], "ratio96": 1.9147373370903351, "ratio_final": 1.914785202186186, "diag": {"ratio": 1.914785202186186, "nonnormal": 0.6440140789038103, "eig_gaps": [0.06344635049218188, 0.5066630190824458, 0.536269467265904], "discness": 0.11937327871219945, "peaks": 3, "deg": 3, "diamW": 4.02659816689329}} +{"family": "random", "seed": 400006, "x": [0.47863329172096325, 0.48793596154177526, 0.6372176039099714, 0.0284562835628297, -0.8836849554439321, -0.16841313139270264, -0.0022889081007685477, 0.3241012857486631, -0.45636832006141087, 1.2309031502173364, -0.7240709123722814, -2.009383641295884, -0.46215936889684694, 0.1655222153944077, -2.214298982073867, 0.8310458265607209, 0.10357192200359389, 0.04631930232135177, 0.07213673277293825, -0.04344062708867509], "ratio96": 1.9594436350872872, "ratio_final": 1.9603417569805628, "diag": {"ratio": 1.9603417569805628, "nonnormal": 1.2692905141439979, "eig_gaps": [0.16746401471759936, 0.5209808721524615, 0.5283556065827749], "discness": 0.05416115878220803, "peaks": 4, "deg": 3, "diamW": 2.902574031158226}} +{"family": "jordan3", "seed": 401006, "x": [0.4777136052106101, 0.35248140929494287, -0.06854745245041499, 0.08456361250332894, -0.47559179222976194, -0.43222931975513346, 1.2334841037062203, -0.4902254144574247, 0.07798739444500802, 0.04137142063186865, 1.355231026190787, -0.04068449073975561, -0.061105423358001724, -0.20281475193284015, 0.0521339665120317, 0.056329094466752355, 1.1130677860472218, -0.09034012277684, 0.025158776997299598, 0.05815317951772048], "ratio96": 1.9787425436967245, "ratio_final": 1.9787114916978592, "diag": {"ratio": 1.9787114916978592, "nonnormal": 0.6969931859501752, "eig_gaps": [0.2685455033717608, 0.29035836434580736, 0.5449834719264761], "discness": 0.06219869401728481, "peaks": 4, "deg": 3, "diamW": 2.265631266900586}} +{"family": "jordan2", "seed": 402006, "x": [-0.08890400897663925, -0.2624976354787355, -0.3779075569854964, 0.012058282264500973, -0.4037853084763352, 0.21474625495157695, 1.2670627624410689, 0.016479592820448002, -0.11605418925057426, -0.686226135665413, -0.13542610034458946, -0.046874905699384545, 0.2721266519153368, 0.11284423036045615, 1.0291075437552584, -0.1753866082358868, -0.007617948230046349, 0.028664230824509015, -0.00458893631260931, -0.030389275996942952], "ratio96": 1.99160500015283, "ratio_final": 1.9916104649078281, "diag": {"ratio": 1.9916104649078281, "nonnormal": 1.210055576623942, "eig_gaps": [0.13649143451114973, 0.26627708795018706, 0.381928253988383], "discness": 0.015489242277681945, "peaks": 3, "deg": 3, "diamW": 1.4970407416472709}} +{"family": "nearnormal", "seed": 403006, "x": [-0.7659576633390519, 0.38975503508264864, -1.1065843864571647, 0.14879262496955314, -1.2335387214227997, -0.4383193871706421, -1.1198160147188005, -0.39220251285054286, -0.0683964730898167, -0.23225867953575263, 0.7468546987946749, 0.5397924054122228, 2.564436238746241, -5.018324259456307, 6.044052512787314, -7.186533974513433, 4.23343591797086, -1.197714964198174, 0.9565975415735437, 1.4323610757267335], "ratio96": 1.9575612428068163, "ratio_final": 1.9605124786432069, "diag": {"ratio": 1.9605124786432069, "nonnormal": 0.34992695012715297, "eig_gaps": [0.233109225988691, 0.33559639490033943, 0.5312989280678324], "discness": 46167255276.48359, "peaks": 3, "deg": 3, "diamW": 1.789891729246906}} +{"family": "companion", "seed": 404006, "x": [0.6781380566474811, -0.7972147023034903, -1.572400798866932, 0.4741713429775596, -1.6618896277407154, 1.1287371524244219, 1.1810548390397682, -0.09682037845502535, 0.4569680111698647, -0.5678805894215304, 1.9246397311918149, 0.664884313606311, 0.6085245285365808, -0.25808124339802985, -0.7389591255302541, -0.3199681377431427, -0.5180025955110048, -0.592395850810844, 0.0215185141308544, -0.10230995807143764], "ratio96": 1.780125726519464, "ratio_final": 1.8022693302847397, "diag": {"ratio": 1.8022693302847397, "nonnormal": 0.5964756427411668, "eig_gaps": [0.16157436135687997, 0.6321642681671328, 0.7412043619293848], "discness": 0.4182798663194334, "peaks": 4, "deg": 3, "diamW": 4.0888585734440515}} +{"family": "peaks", "seed": 405006, "x": [-1.1419118386480398, -0.40955671665534976, -0.1721822826431238, -0.07778192444747811, 0.16137979685811044, 0.22121779785775608, 1.0463878612064088, 1.021243806466734, -0.2620039701378799, -0.03840032138143459, 0.12363797089089637, 0.0908202951125826, -1.4328465563240305, -2.5035457620609467, -2.4527590597342175, -1.8428283731440982, 1.5207386894886987, 0.8572447715455168, 0.6951247435177166, 0.08664003621127568], "ratio96": 1.9152487471896869, "ratio_final": 1.9212837509770775, "diag": {"ratio": 1.9212837509770775, "nonnormal": 0.9949379766720013, "eig_gaps": [0.2379285442671748, 0.5443759829055729, 0.7690473422151876], "discness": 0.4352479611365115, "peaks": 5, "deg": 3, "diamW": 1.8827329381893674}} +{"family": "real", "seed": 406006, "x": [0.20028678957252943, 0.04987978972969981, 0.31328807148201576, 0.01846701689458702, 0.17800085797955956, 0.04435230193614961, 0.3191462580895724, -0.007836581848676596, -0.27672953846139803, -0.056185470385430425, -0.20248183849878898, -0.02565545274375014, -0.5740675909760251, -0.15199523572124785, 1.8934955191926033, 0.21795612371750883, 0.25283256690140277, 0.33081095676531436, 0.4651695086025137, -0.06944594864961291], "ratio96": 1.974460847202167, "ratio_final": 1.9763202426022828, "diag": {"ratio": 1.9763202426022828, "nonnormal": 0.7837545030619908, "eig_gaps": [0.04822108625623177, 0.24631431777376378, 0.2892724487887415], "discness": 1401087496.616828, "peaks": 3, "deg": 3, "diamW": 0.4761647559737326}} +{"family": "random", "seed": 400007, "x": [-1.7271974843956723, -1.3307173372291028, 0.9342054411926182, -1.6072672721446133, -2.326967614663957, 1.7583782074919152, 2.6354734332322716, -1.282194938696977, 0.22886904940067898, -1.3560563321396204, 2.4227444424658984, 1.4369103261412157, -3.710980836557621, 13.91049036141306, 3.5479043452632757, 5.266578910299557, -0.8885590888750532, 3.1066713889548545, -2.732535811112286, -1.2251635406336698], "ratio96": 1.47935934464597, "ratio_final": 1.4825509958198984, "diag": {"ratio": 1.4825509958198984, "nonnormal": 0.7179693033103951, "eig_gaps": [0.42039853141742434, 0.49440754732533504, 0.7363130703655532], "discness": 0.1582060696059202, "peaks": 3, "deg": 3, "diamW": 6.364752691845483}} +{"family": "jordan3", "seed": 401007, "x": [-0.23990780783593102, -0.45277256917452346, 0.035258983070475064, 0.40010524004005, 0.22276981934322648, 0.32671683741842095, 1.252562316244276, -0.05290650274406739, -0.42173326953585777, 0.35833500655730477, 0.9667544482577632, 0.5622353513558098, 0.0185843151238841, -0.049796182909323534, -0.12829691892727413, -0.3865614152633883, 1.2520341909045287, -0.280135177653246, 0.1110900276544754, -0.4086311989177509], "ratio96": 1.978459560993487, "ratio_final": 1.9784330459657076, "diag": {"ratio": 1.9784330459657076, "nonnormal": 0.7693817038224604, "eig_gaps": [0.09970809729324441, 0.4437567775179032, 0.44885426947901735], "discness": 0.0470015421385097, "peaks": 4, "deg": 3, "diamW": 2.0195029734736525}} +{"family": "jordan2", "seed": 402007, "x": [0.26665838195127134, 0.0640304725161513, 0.16347612139243972, 0.12743623276967211, 0.18638742679347298, 0.146708482375494, 0.28023340580960826, 0.26874593291937754, -0.36574668595684606, -0.20622109820791154, 0.017300261346530968, 0.056087999820363825, -0.31719101950103934, -0.08482792471879236, 1.4484895367671293, -0.1559848125664955, 0.03923994409868693, 0.33784815332331863, -0.1609793261886948, -0.10474018452002712], "ratio96": 1.9921684385032334, "ratio_final": 1.9921654931823096, "diag": {"ratio": 1.9921654931823096, "nonnormal": 0.9352248673993476, "eig_gaps": [0.05126140676795281, 0.19730418455475635, 0.20735810777389252], "discness": 0.03682290617984322, "peaks": 3, "deg": 3, "diamW": 0.5840468557943083}} +{"family": "nearnormal", "seed": 403007, "x": [-2.0958263771046246, 0.9701023708898655, 0.6164119147489941, 0.7756431918705813, 0.17083768167684252, -0.9791742802419883, -3.5641232463507273e-06, 6.772032910948186e-06, 1.1496464631278123e-05, -2.9272011800299707e-06, -0.356252371217637, -0.2690348562352274, -0.16799366373451458, 0.30846891680998456, -0.008491575676104483, -0.010693331105547498, 0.1930001992003953, -0.7225364008953097, 1.0662043126999898, 0.1764901894661851], "ratio96": 0.999999999977969, "ratio_final": 0.9999999999855319, "diag": {"ratio": 0.9999999999855319, "nonnormal": 0.1569196375646732, "eig_gaps": [0.593760806071206, 0.891771358993788, 0.9804354427669555], "discness": 0.4904988207744711, "peaks": 1, "deg": 3, "diamW": 3.049212550471817}} +{"family": "companion", "seed": 404007, "x": [-0.07935847544128258, 0.1696381009303255, 1.1352124039824503, -0.15961559778566276, -0.39430256442774175, 0.5081402967911511, 1.8969489142449998, 0.09283326883950821, -1.1493055953530875, -0.7118155423153181, 1.1203721797830306, 1.3025041829869053, -2.3073958249794297, -0.43826862531356037, 0.36747562660831046, -2.0319113383563234, 0.968562569523401, 3.6915163677134735, 0.5861255011108146, 0.3171651183809855], "ratio96": 1.9511460663082802, "ratio_final": 1.9513804021932275, "diag": {"ratio": 1.9513804021932275, "nonnormal": 0.8012855960238126, "eig_gaps": [0.13706704183030086, 0.37305900094019234, 0.494758391595765], "discness": 0.06800897941042856, "peaks": 4, "deg": 3, "diamW": 3.373214122922498}} +{"family": "peaks", "seed": 405007, "x": [-0.973207552073375, -1.4253653104832993, -0.26138240654902, 0.08474461463385796, 0.09132457192601559, 1.089651387682911, 2.314842792736478, 0.36220712431232255, 0.22948750746941293, 0.3109728179097097, 1.9483298913820557, -0.2379630867976427, -0.4562084223261708, -2.666478950685109, 1.053158648790393, -1.4215321117877782, 1.368100920932, -2.6405224726413254, -0.23838834900741335, -0.15145099907731274], "ratio96": 1.911489928763168, "ratio_final": 1.9324690744761903, "diag": {"ratio": 1.9324690744761903, "nonnormal": 0.7041259506728255, "eig_gaps": [0.25882706788689464, 0.4057286030054735, 0.6637187510134418], "discness": 0.1332300150112916, "peaks": 4, "deg": 3, "diamW": 4.114743382705791}} +{"family": "real", "seed": 406007, "x": [0.4232805006186566, 0.3032863466427486, 0.1079836398073164, 0.36446421411755003, 0.2902826232154063, 0.15080706600441307, 0.05382965082075028, -0.07013063794574523, 0.2464980655309848, 0.09515916196395252, -1.0812407159761497, -1.121832427984087, -2.0454676493970085, -1.1928086162389375, 6.426534692902552, -2.9185576933211994, -0.1788615215363415, -0.05543002642855899, -0.01820183772369944, -0.07788665355238644], "ratio96": 1.9978320697756538, "ratio_final": 1.998177627046962, "diag": {"ratio": 1.998177627046962, "nonnormal": 1.1867260636607906, "eig_gaps": [0.1259377741763055, 0.17481573930682828, 0.1999104827018917], "discness": 0.0088784670732251, "peaks": 3, "deg": 3, "diamW": 1.6066054693574767}} +{"family": "random", "seed": 400008, "x": [0.3663403240379718, -1.1630356043286478, 1.2107192498068509, 0.3936918947594209, -0.038502087020308225, 0.39526120009010735, 1.1705550787976575, 1.8976072804933066, -0.06269606842011673, -1.2818891589261967, 2.0495081609776005, 1.555995408049099, 0.18351717531629252, -1.820402701816372, -1.11558274367599, 2.124843466657353, 1.3520182725911765, -1.3981797771030908, 0.28241729181316977, -0.05694894795418612], "ratio96": 1.9677583889649517, "ratio_final": 1.971935983053439, "diag": {"ratio": 1.971935983053439, "nonnormal": 0.7316288357018575, "eig_gaps": [0.3066673255202305, 0.3952399864982682, 0.4347522613966448], "discness": 0.03423094526981491, "peaks": 4, "deg": 3, "diamW": 4.073542299980285}} +{"family": "jordan3", "seed": 401008, "x": [0.27264417516194184, -0.40355358427079496, -0.7706821774908218, -0.18341435111128662, -0.23048979941639292, -0.11135955917322606, 1.2087084730597066, -0.42294884578635816, 0.17915083226326187, -0.20651481397510119, 0.9728925904740653, -1.0830945158741114, -0.24414555321492099, 0.202622436898433, 0.32658301662328304, 0.9770432965626488, 1.744250967534314, 0.7485854415286339, 0.3363813613813472, 0.029639818913553655], "ratio96": 1.9849440399298166, "ratio_final": 1.9851595972644236, "diag": {"ratio": 1.9851595972644236, "nonnormal": 0.6776964068542172, "eig_gaps": [0.2453687395485127, 0.2619594851466761, 0.480086807418659], "discness": 0.06793844681529179, "peaks": 4, "deg": 3, "diamW": 2.221052233486238}} +{"family": "jordan2", "seed": 402008, "x": [0.3519641992289182, -0.04579720347873614, -0.09893467737384185, -0.008689455604605364, 0.4718924716103322, -0.08844095784534273, 0.8307727729276292, -0.24756938141068757, 0.25568642097485117, 0.2574770640178582, -0.19787738215038045, -0.2543383852506127, -0.17870936527071085, -0.0570355804312576, 1.349151372773137, 0.7681295252837211, 0.0756724444141999, 0.028864119031639504, -0.18885460152250347, -0.1428852586742786], "ratio96": 1.9832085194658557, "ratio_final": 1.9832604521619137, "diag": {"ratio": 1.9832604521619137, "nonnormal": 1.1150821174767303, "eig_gaps": [0.11798036972872193, 0.41935322170949074, 0.5342413163740265], "discness": 0.038059459484859864, "peaks": 4, "deg": 3, "diamW": 1.0788595620742354}} +{"family": "nearnormal", "seed": 403008, "x": [-1.2136214967580576, -0.5142285823882673, -0.1465749602247652, -0.06667532011424951, -1.4708485482677087, -0.6872232275816879, 0.008563858765450022, -0.004461839829838391, -0.5851578828624939, -0.3969130031179755, -0.011592313923957662, 0.004905244349369347, 1.0739739406249045, 0.8834791741412173, 0.29789187133457556, 1.8413012504767763, 1.8196464476688146, 1.050587273626323, 0.9967483982229499, -0.44381995529477514], "ratio96": 1.7180459040155245, "ratio_final": 1.6624134520100349, "diag": {"ratio": 1.6624134520100349, "nonnormal": 0.1577304130451886, "eig_gaps": [0.1829134951908633, 0.6827675971972664, 0.8629451110530018], "discness": 189183537176.33398, "peaks": 1, "deg": 3, "diamW": 1.6947276261115944}} +{"family": "companion", "seed": 404008, "x": [2.664009232228862, -0.3456109942432075, 4.972441164947222, 0.14076840622208842, -0.962158746555563, -0.024445359725746282, 5.072146651714034, 0.1659179765008987, -3.0830970186520457, -1.4347715708696072, 4.446036622431034, 2.825890644636497, -1.5096694781141298, 0.5443261783400037, 1.4797321859270802, -3.755333038337281, -0.3871979566872791, 0.6381976898419595, 0.035510502723660964, 0.046969482821620556], "ratio96": 1.899294501088215, "ratio_final": 1.9150984233362012, "diag": {"ratio": 1.9150984233362012, "nonnormal": 0.6237362157721753, "eig_gaps": [0.2364953070268628, 0.3649371813749639, 0.595159149041349], "discness": 0.127041889561606, "peaks": 4, "deg": 3, "diamW": 9.97531361954089}} +{"family": "peaks", "seed": 405008, "x": [-1.1314193435648412, -1.00250572596297, 0.11063819796007958, 1.2577304216756229, 1.27383486195379, -0.3335212297564938, 2.2979015244603787, -0.39986638719297934, 0.12205146689797608, 0.39629482333413946, 2.020309128553821, -1.2887045489219133, 2.995757357409451, 0.3250184146572523, -0.8128331833086984, -1.1728612113036694, -2.5749123866084087, 1.7710169154296285, -0.37568249439298873, -0.48005946119030796], "ratio96": 1.9000558645184835, "ratio_final": 1.9005921664070224, "diag": {"ratio": 1.9005921664070224, "nonnormal": 0.7383390533156136, "eig_gaps": [0.4562802022192868, 0.5779246784507605, 0.5970157071751007], "discness": 0.050636755992918676, "peaks": 4, "deg": 3, "diamW": 4.31986265628844}} +{"family": "real", "seed": 406008, "x": [0.659656126983944, 0.15189431804171882, -0.4703473851413998, 0.15994728395246607, -0.5600274361703954, -0.8525039622187867, 1.3899332166086935, -1.6762096475073651, 1.0697186474376594, -0.10308842923344212, -0.2797099957286665, 0.6333234098019448, -0.978903263955067, -1.1646340066165992, 1.9510508922728858, 1.2181426055373, 0.5070068044118878, 0.365328149243664, 0.12117339301094164, 0.07734093893904068], "ratio96": 1.7606056140911168, "ratio_final": 1.7689261204340256, "diag": {"ratio": 1.7689261204340256, "nonnormal": 1.1700312175947936, "eig_gaps": [0.36620284600231395, 0.4071377270911714, 0.56926100085038], "discness": 0.03279861213723839, "peaks": 3, "deg": 3, "diamW": 2.775552672435289}} +{"family": "random", "seed": 400009, "x": [-0.22774092893790934, -0.1216641819292445, -0.4204469161047361, 0.7866393496535915, 0.288704211557278, -0.45537575151179077, 0.05322850940166426, -3.4211980300379783, 0.36298019870612186, 0.40746809888878244, 2.0735396358383764, 2.5748028511991663, 0.2701768202932438, -0.2893503324187542, -0.1646704355750226, -0.32230525695657564, -0.045098664957112185, -1.765473364603786, 0.03362495212871314, -0.05559961284806482], "ratio96": 1.9946504192233194, "ratio_final": 1.9946369355893576, "diag": {"ratio": 1.9946369355893576, "nonnormal": 0.7129899058436834, "eig_gaps": [0.12305906438616224, 0.1858292087563823, 0.28623434277059073], "discness": 0.015763486225498797, "peaks": 3, "deg": 3, "diamW": 4.996635462866935}} +{"family": "jordan3", "seed": 401009, "x": [0.03854468427709147, -0.21022928805580704, 0.0506578546619329, 0.38943210640504344, 0.09142038511193193, -0.13194916233997026, 1.2970898211924482, -0.18152656507851092, -0.0215318444762966, 0.15296657651060486, 1.2721598273369157, 0.407468069685303, 0.04422913860446514, 0.008915778283282806, -0.13939731143541584, 0.024819265834887878, 1.366880516446983, 0.04509611587755551, -0.041354312861967266, 0.22256727134678608], "ratio96": 1.9947657274448514, "ratio_final": 1.9947499524268297, "diag": {"ratio": 1.9947499524268297, "nonnormal": 0.728721986424458, "eig_gaps": [0.04775604062686137, 0.26438481408176895, 0.3032162756641244], "discness": 0.020291685796728715, "peaks": 3, "deg": 3, "diamW": 1.978072330693802}} +{"family": "jordan2", "seed": 402009, "x": [0.22183406818414309, -0.3321832194025729, -0.40849962038569954, -0.06922118732437961, 0.278497682336157, 0.09980074439975051, 1.2826594794200945, -0.10596219242046855, -1.6563503834895157, 0.42080913091138317, 0.9945353503740935, 0.050362417825558975, 0.026560965590623903, 0.3714469160903635, 2.040100655174465, 0.8938034619853016, 0.042665425660646364, -0.012651063073998527, -0.007046327805114161, -0.017651152478195452], "ratio96": 1.9902312035846754, "ratio_final": 1.9904085835769856, "diag": {"ratio": 1.9904085835769856, "nonnormal": 1.3661241825453168, "eig_gaps": [0.17799973732628177, 0.27903520714674857, 0.2890440277791622], "discness": 0.011276716001756507, "peaks": 3, "deg": 3, "diamW": 2.4476688691882047}} +{"family": "nearnormal", "seed": 403009, "x": [-2.261226861468677, -0.3003322336238823, -0.022968297858582024, -0.420352629746012, 1.5824849181249516, 0.3742987391553157, 0.8796882768131438, -1.7375613588103362, 1.4582913741048849, -0.6289426982009182, -1.9070421162403446, 1.5427567451418556, -2.738006525376435, 4.990450730579992, 0.06665217112313647, -1.6327440351293863, 1.1137290167196339, -2.4593881059189684, 0.36750743242940986, 0.13791377050920234], "ratio96": 1.7749212299156645, "ratio_final": 1.7786737363778258, "diag": {"ratio": 1.7786737363778258, "nonnormal": 0.836273567732578, "eig_gaps": [0.3408617009768091, 0.4265111679550962, 0.7425674253152206], "discness": 0.17031955619375466, "peaks": 3, "deg": 3, "diamW": 5.255370321655844}} +{"family": "companion", "seed": 404009, "x": [-0.47953131525842196, 1.0135699649509924, 1.1274456053034565, 0.8229293234116888, 0.447596741949128, -0.9545088697078681, 1.7048672075918638, -0.05689587655826359, -1.4568181404402187, 0.5310104002156508, 1.6189119590903311, 0.5436015078068784, -0.9395735625339497, 0.8987905473702422, 0.1402145709363042, -1.9167010744421695, 0.7546786102828538, 1.8287537229948052, 0.2505592572055595, 0.20412091873636218], "ratio96": 1.7923898947444397, "ratio_final": 1.795200320138665, "diag": {"ratio": 1.795200320138665, "nonnormal": 0.7703200637539882, "eig_gaps": [0.448773949957648, 0.5277476674972144, 0.6033188776751638], "discness": 0.05210113169966838, "peaks": 3, "deg": 3, "diamW": 3.6059257795764674}} +{"family": "peaks", "seed": 405009, "x": [-0.7068786440132873, -0.09410910025856656, -0.08561227921145037, 0.37625518561316074, -0.1519356983949856, -1.091822262573443, 2.3215219516527306, -0.48160579471035286, 0.34479439496156417, 0.4226598359761319, 2.105438267215958, -0.6627049716399123, 0.5387849201432447, 2.0015857003613235, 1.495419116298284, 3.9930089678493683, 3.562255546932391, 1.8349674222382366, -0.47608679349979105, -0.035811581855137074], "ratio96": 1.9658277206488137, "ratio_final": 1.9657348708525542, "diag": {"ratio": 1.9657348708525542, "nonnormal": 0.6919349364269667, "eig_gaps": [0.21575333661915772, 0.3160997840111913, 0.4068910854583067], "discness": 0.03404861946510586, "peaks": 3, "deg": 3, "diamW": 3.6117154824153883}} +{"family": "real", "seed": 406009, "x": [-0.37284670761904504, 0.0293887097336996, -0.018522783882466948, 0.006435549263959043, -0.3373520703591677, 0.004326600233458357, 0.4147470295667798, 0.03716035756059122, 1.3144811487381274, -0.010562581081042847, 1.3253691352061527, 0.054628003139014475, 0.4580649006121915, -0.08106992302865976, 2.8900650306552955, 0.03209881855142008, 0.009313055527221059, -0.016734127394161094, -0.01619871494432045, -0.007460463057680774], "ratio96": 1.998642770727212, "ratio_final": 1.9986423314570214, "diag": {"ratio": 1.9986423314570214, "nonnormal": 1.3348378307665354, "eig_gaps": [0.022462406332965982, 0.1648258988376734, 0.18355556940933307], "discness": 0.004919149520888344, "peaks": 3, "deg": 3, "diamW": 1.934382057946099}} +{"family": "random", "seed": 400010, "x": [4.856639551490209, 3.6646571690516847, 2.8745685127335445, -1.9338118723509474, -1.6671272273111688, 3.694518307049382, 1.4208738587602596, -4.658863316165691, -1.6523013114433969, -3.229457436235249, -3.746301993855038, 6.876943177885096, 11.390685949573218, 5.156109842632173, -10.657682300350091, -8.20424004032496, 0.8686878220323329, -1.747136165628734, 0.7089508389620098, -0.2525604106082646], "ratio96": 1.4880523186988779, "ratio_final": 1.4882500726649115, "diag": {"ratio": 1.4882500726649115, "nonnormal": 0.7863933697500112, "eig_gaps": [0.4858696563107541, 0.5337169895193274, 0.5916707133714624], "discness": 0.0556075991695392, "peaks": 3, "deg": 3, "diamW": 12.223397883181907}} +{"family": "jordan3", "seed": 401010, "x": [-0.3149361587408003, -0.21551196733305505, 0.339050491458098, 0.6056176203228842, 0.07777994851851322, -0.44507405430720837, 1.1161723718165852, -0.5003681068699157, -0.09718400664794431, 0.8295971120324027, 1.3301641032656117, 0.32579238355274726, -0.09525119994046503, -0.21063944441270524, -0.08260118119993072, -0.05262079689801061, 0.8464161957684946, -0.9464314057087773, -0.2535401813273302, -0.055805525021297694], "ratio96": 1.9638052265768613, "ratio_final": 1.9638134664520799, "diag": {"ratio": 1.9638134664520799, "nonnormal": 0.7285249786253036, "eig_gaps": [0.19389124821095202, 0.4474386051853289, 0.4614830330634066], "discness": 0.05053047081210302, "peaks": 4, "deg": 3, "diamW": 2.3461077434622393}} +{"family": "jordan2", "seed": 402010, "x": [-0.4705287650038279, 0.42166434140105413, -0.1718318218675568, 0.4510703707222654, 0.49956016229896766, 0.4543597439405777, 0.8746949028950749, 1.0023744599041588e-05, -0.0035285305848654328, 0.004626425329218299, 0.005019851121867839, 0.0008216205316423008, 0.19711407940523973, -0.6161875452120507, 0.997986660860545, -0.06507005689542798, -0.23505349471640766, 0.5341524006395462, -0.1820064684907137, 0.3809988618241933], "ratio96": 1.872399103464923, "ratio_final": 1.8562662258619247, "diag": {"ratio": 1.8562662258619247, "nonnormal": 0.6172643776303153, "eig_gaps": [0.2338984695993838, 0.5232190152823144, 0.7564151627318185], "discness": 1.8301968914279765, "peaks": 4, "deg": 3, "diamW": 1.283210323738045}} +{"family": "nearnormal", "seed": 403010, "x": [-0.9610754374302735, -0.6719738245615725, -0.7857233904447865, -0.5497906539944826, -1.1400208942459042, -0.8957377802591684, -0.02442360831907832, -0.0767926443835779, 0.7139188044994125, 0.00374520667182384, 0.05566666916645961, -0.12969757516720723, -1.7510823291064395, 0.4279630895602277, 1.0259909544685493, 3.0928670423848805, 2.69394800443116, 1.8801664219633998, 1.093431860446832, 0.49782049463338446], "ratio96": 1.8408427699229664, "ratio_final": 1.8421403904378362, "diag": {"ratio": 1.8421403904378362, "nonnormal": 0.16338197271246674, "eig_gaps": [0.2687331934158094, 0.36026498458146017, 0.6226412251172806], "discness": 11422575498.476671, "peaks": 3, "deg": 3, "diamW": 0.7952941617792536}} +{"family": "companion", "seed": 404010, "x": [-0.23358967559571897, -1.28090873738949, -0.7468567890202502, 1.3670106960780437, -0.20804909800634963, 1.9811735913249477, 2.4923266400198605, -0.7965237822350789, 0.001541559998433733, 0.48725428175651775, 2.6326447064789273, -0.11277816185833346, -1.4768915993350487, 1.1615159123621241, -0.06448308105005324, 2.080310714376713, -1.261041107460945, 0.8584670953106981, -0.08914025674475612, 0.05629974699860085], "ratio96": 1.92877613632646, "ratio_final": 1.9287354498093967, "diag": {"ratio": 1.9287354498093967, "nonnormal": 0.6416282133399552, "eig_gaps": [0.16037269234467313, 0.5294387506893278, 0.6403388167155266], "discness": 0.1629769584946493, "peaks": 4, "deg": 3, "diamW": 5.0944628490555015}} +{"family": "peaks", "seed": 405010, "x": [0.24024679289498257, -0.09731390730206357, 0.03210074607578857, 0.8053772392538259, 0.8965040750705746, -0.09321264765640486, 1.2196734180511108, -0.44142246410435737, -1.1773530493057107, -1.8214142335284798, 2.04247424093146, 0.8308718314546747, 0.1968334565743245, 1.5307461436804362, -4.0126503716519135, -1.0851022404481183, 0.2776240211350127, -0.004356946561595831, -0.016512687924860693, 0.02821681622848994], "ratio96": 1.9707837571591873, "ratio_final": 1.9707916175806544, "diag": {"ratio": 1.9707916175806544, "nonnormal": 1.3046358344313416, "eig_gaps": [0.18618126605255556, 0.26280980637326545, 0.35372898041396844], "discness": 0.019388377633169262, "peaks": 3, "deg": 3, "diamW": 3.524898671668976}} +{"family": "real", "seed": 406010, "x": [-0.13682990206278445, 0.3913885021690906, -0.12074141731529525, -0.07047287059811458, 0.13528976648032126, -0.02216296556349686, 0.188846236127049, 0.06948991252943004, -0.16271124501777967, 0.4279605114350058, 1.206258133411756, 0.2073399997472349, 0.2083695558312193, -0.15661107949724834, -2.9246889293417375, 0.48414049450203656, -0.4418169137246779, 0.1189526676206738, -0.017677713428430462, -0.00890760876699348], "ratio96": 1.9787685151411023, "ratio_final": 1.9787659977707426, "diag": {"ratio": 1.9787659977707426, "nonnormal": 1.2797728668030108, "eig_gaps": [0.19336734652642415, 0.34297983996117887, 0.3674025622666667], "discness": 0.0068932916331201095, "peaks": 4, "deg": 3, "diamW": 1.347430511092709}} +{"family": "random", "seed": 400011, "x": [6.684187815291834, -8.914533312051766, -7.443576013534653, 2.361941844938748, 7.354335432749464, 9.295295938199336, -4.33093454663995, 30.381515223324296, -4.576483796722105, -5.200174912813532, -34.38577199236677, 1.0975426904637107, 1.4566245048452813, -5.063322773548105, -20.10264502690636, -6.5022029921920055, 53.10871810533355, -3.2517042157857032, -0.10758533057022972, 0.3926349548519319], "ratio96": 1.9019796032179987, "ratio_final": 1.9121754407897367, "diag": {"ratio": 1.9121754407897367, "nonnormal": 0.7552378044154928, "eig_gaps": [0.32786444476236953, 0.36266688210669373, 0.36559322379457593], "discness": 0.008152510539096495, "peaks": 3, "deg": 3, "diamW": 49.842707918740025}} +{"family": "jordan3", "seed": 401011, "x": [0.21351911923788608, 0.1506909270714713, -0.1768068763991678, -0.26586436515096806, 0.19310643202758504, 0.41838348010739074, 0.9157000619058213, -1.007008298246707, -0.3218669803707857, 1.1988710685927915, 0.5408501367932514, 0.4585008897582016, 0.10595034392119682, -0.07458772262724345, -1.4050802608772703, -0.7432569106861454, 0.017215053345738224, -0.034941175086837584, 0.014330583854682551, -0.03521719726700879], "ratio96": 1.9921677503216075, "ratio_final": 1.9921639591736189, "diag": {"ratio": 1.9921639591736189, "nonnormal": 1.3441349897745958, "eig_gaps": [0.12964834997104202, 0.2756739310870979, 0.37563021273762703], "discness": 0.01813847966604149, "peaks": 4, "deg": 3, "diamW": 2.0707529389739596}} diff --git a/problems/crouzeix/data/census_i5.jsonl b/problems/crouzeix/data/census_i5.jsonl new file mode 100644 index 0000000..5b0e0df --- /dev/null +++ b/problems/crouzeix/data/census_i5.jsonl @@ -0,0 +1,83 @@ +{"family": "random", "seed": 500000, "x": [2.2010809995628335, -1.9464386842726633, -1.2698301666292287, -0.89428803019729, 1.3610875358971162, 1.7102617018999196, -2.8252671005099472, 2.5412514244086974, 1.8303891128175163, 0.9336926885547425, 0.9446971997773101, -3.7558848154159383, 0.20061937057321858, -0.48227195365050324, -1.6462906776973192, -1.0902350573899942, 0.31713912319027393, 1.0154481475251607, -0.0777387395391915, -0.018374794815367405], "ratio96": 1.8253779590675354, "ratio_final": 1.8845616755755084, "diag": {"ratio": 1.8845616755755084, "nonnormal": 0.689697639320076, "eig_gaps": [0.5258708856485272, 0.5367748155103692, 0.5440038018508758], "discness": 0.0283601437991864, "peaks": 3, "deg": 3, "diamW": 6.896898606843152}} +{"family": "jordan3", "seed": 501000, "x": [-0.09798776276255913, -0.17362165051916933, 0.40148514812217984, 0.0713509880183586, -0.26531901460044605, -0.22642863415543446, 1.4657720102721816, 0.26626103947855356, 1.152503973526167, -0.10149419229378527, 0.8898035635065351, 0.012078541125608914, -0.27692657742021753, 0.17413333191125924, 1.514521996616328, -0.7713745301452021, -0.06509784928808299, -0.03800720597360517, 0.011888901907936865, 0.007039782768977668], "ratio96": 1.9786970011397063, "ratio_final": 1.979742431611181, "diag": {"ratio": 1.979742431611181, "nonnormal": 1.3073651285101924, "eig_gaps": [0.08293634463974697, 0.26294898946821815, 0.3451738207769734], "discness": 0.00971227306109078, "peaks": 3, "deg": 3, "diamW": 2.115671050451388}} +{"family": "jordan2", "seed": 502000, "x": [-0.07297242819679475, -1.1017837014494511, -0.007237127647621706, -0.1556690019805843, -0.39838040389296564, 0.2225268261918069, 1.749410684025755, 0.2920917566448649, -0.06084562506236932, -0.18798559834238168, -0.10934705833463322, -0.053172035158510794, 0.2537801433106527, 0.8864028531664594, 1.259477795852082, -0.2968643529849976, 0.0454469791824586, 0.23305786443657373, 0.10118835091616929, -0.007421163893165838], "ratio96": 1.9722743730220806, "ratio_final": 1.9725207646608844, "diag": {"ratio": 1.9725207646608844, "nonnormal": 1.096524119175018, "eig_gaps": [0.2679904453911002, 0.4671372758308442, 0.6716996727205548], "discness": 0.13265552335264527, "peaks": 4, "deg": 3, "diamW": 2.030228837510328}} +{"family": "nearnormal", "seed": 503000, "x": [0.9514264593789671, -2.347125623512057, -0.8134464636237098, -0.8969761677742097, 0.6396030450580132, 0.541094763399647, -2.6685891043895647e-07, -2.722830965954275e-06, 1.3735402977143419e-06, -1.1667995633029469e-05, 0.13387084833551727, 0.0482949758842278, -0.02068051617361558, -0.2824950875162474, -1.0302531355781037, -1.187731093383067, -0.9359994321335581, 1.700695992494631, 1.1830909467027557, 0.10284111740736392], "ratio96": 0.9999999999811986, "ratio_final": 0.9999999999897815, "diag": {"ratio": 0.9999999999897815, "nonnormal": 0.0479452208509749, "eig_gaps": [0.7027770289696583, 0.7852365503923593, 0.9986362912686834], "discness": 1.6661929448049169, "peaks": 1, "deg": 3, "diamW": 2.9089714381264984}} +{"family": "companion", "seed": 504000, "x": [0.9775374278349032, 1.233439622907949, 0.46796876025924433, -0.37387983350426407, 0.8599935382002286, 1.0419597960369855, 0.0006866978905461961, 0.0007649866345611985, 0.12266955355014929, -0.6992793978376874, -0.0003775349167190945, -0.0008362368203637133, 0.7776427822991061, -0.28397222075708883, 0.7361749974675182, -0.7845477438405442, -0.7477518662234235, 0.8274827940607006, 0.4850179878508275, -0.4734185788720944], "ratio96": 1.7900755250285851, "ratio_final": 1.7905543163271889, "diag": {"ratio": 1.7905543163271889, "nonnormal": 0.1447556930687938, "eig_gaps": [0.11530365285323353, 0.7539336519151942, 0.8653215624718797], "discness": 230979510196.4383, "peaks": 3, "deg": 3, "diamW": 1.9485937092487953}} +{"family": "peaks", "seed": 505000, "x": [-1.0606376079384545, 0.45599240756357196, 1.1951877362739876, -0.18073502213640832, 1.1539411157667614, -0.16273511406503516, 1.3692730439940373, 0.8384668224592727, -0.09011065039352059, -0.6176611892487054, 1.5564899012580486, 0.40480261342323703, -1.0652268987585274, 0.9051772403391852, -1.3177682708245633, 0.1621699844562733, 1.7433215618307192, -0.4991079651114599, 0.08341153388957076, 0.16485957998273681], "ratio96": 1.8785919580536365, "ratio_final": 1.8785472999010968, "diag": {"ratio": 1.8785472999010968, "nonnormal": 0.6487491843464325, "eig_gaps": [0.012723597949459922, 0.6500990267993801, 0.6627021766824078], "discness": 0.18383000428868004, "peaks": 4, "deg": 3, "diamW": 3.5369803068350807}} +{"family": "real", "seed": 506000, "x": [1.965312424072731, 0.14450921856776555, -0.7837662964051677, 2.158561598058486, -1.6873611609670132, -2.6394955686864647, 1.7115396318888432, 3.3705297582842606, 2.760559106144206, -0.5709724700319296, -4.335412709790507, -2.312499203761174, -2.2406932866966707, -0.03694865485921059, 2.0871018989102295, -1.6745822718972447, 2.684742672867243, 4.716330459644395, 0.4848585542160312, 0.002513386941721851], "ratio96": 1.893569356397994, "ratio_final": 1.8950095242900347, "diag": {"ratio": 1.8950095242900347, "nonnormal": 0.7762494495310546, "eig_gaps": [0.4074353780503413, 0.5490816903607489, 0.5837195223607875], "discness": 0.04476857709236547, "peaks": 4, "deg": 3, "diamW": 8.364292229525603}} +{"family": "random", "seed": 500001, "x": [-0.32451703623452827, -0.018664267400409797, 0.277264828121787, 0.44945329734371164, 0.05153146962232798, 0.09355566721037682, -1.4437452797936519, 0.3695577566198392, -1.4041161410296854, -0.06531523128994027, 0.035661655312237195, 0.5374024988486452, -0.7035202236342175, -0.40332272610665265, 1.427761317832099, -2.6445800620558955, 0.03772672940714584, -0.08277529345984677, 0.054159864026256894, 0.062406033866259605], "ratio96": 1.9896281567697067, "ratio_final": 1.989647419435718, "diag": {"ratio": 1.989647419435718, "nonnormal": 1.3661119810966214, "eig_gaps": [0.17452549962143868, 0.1874280866260705, 0.33906374830300945], "discness": 0.03013954368771768, "peaks": 4, "deg": 3, "diamW": 2.2485867021911887}} +{"family": "jordan3", "seed": 501001, "x": [-0.06676528698358883, -0.03143590796544164, -0.1862869641929027, -0.15728533665334532, 0.10536666765344921, 0.34613509060450565, 1.252945153974515, -0.18138884355036575, 0.13118638229781088, -0.1771079345626435, 1.163287401364013, -0.3058706140517689, 0.012256852545627674, -0.07440279612666323, 0.10273556270880962, -0.192113039623867, 1.3663520279657484, -0.3874121295125126, -0.09032172719878143, -0.06681052355270811], "ratio96": 1.99531778862005, "ratio_final": 1.9953155484498082, "diag": {"ratio": 1.9953155484498082, "nonnormal": 0.7101638555128271, "eig_gaps": [0.09337658316278655, 0.223248336538452, 0.31301176206977355], "discness": 0.020631702259158665, "peaks": 3, "deg": 3, "diamW": 1.8587235834505667}} +{"family": "jordan2", "seed": 502001, "x": [0.7907785478031975, -0.9009602761318234, 0.5102374267968905, -0.43668580757044073, 0.182582783663714, -0.6664152161258943, 1.2315521710918673, 0.48131274895022175, -0.7395842030181672, 0.05545628501429549, -0.2922977563642658, 0.042636907089796894, -1.030466776080794, 0.7486095714866186, 1.1409348229897498, 0.10043195825595105, 0.0015796338564985565, 0.25865534248810496, 0.04909794121084454, -0.04148089289606957], "ratio96": 1.9835916006758831, "ratio_final": 1.9835924689440168, "diag": {"ratio": 1.9835924689440168, "nonnormal": 0.7376664048897879, "eig_gaps": [0.24513483313107046, 0.33229640730883725, 0.3993140885011319], "discness": 17384863855.749035, "peaks": 4, "deg": 3, "diamW": 1.6324339473517828}} +{"family": "nearnormal", "seed": 503001, "x": [0.16209843070403318, 0.45885606673109447, -0.7183997323539497, 1.5498397045902137, -0.003656868732838745, 0.1712607253475316, 0.2799845426959468, 0.28001111064504497, -0.3160310201274161, 0.33523921268425344, -0.21999066712422746, 0.2600019699593843, -0.6478125224995749, -0.44869525649195846, -0.9379058938590823, -0.712699571399069, 0.23269225077849276, -1.402819016038616, 0.5608558604228999, 0.16911728767135858], "ratio96": 0.9999999999696355, "ratio_final": 1.2480859785227605, "diag": {"ratio": 1.2480859785227605, "nonnormal": 0.3552450807007695, "eig_gaps": [0.18678369980861795, 0.7888872768592726, 0.873785735023703], "discness": 184295262488.01663, "peaks": 2, "deg": 3, "diamW": 1.7771498014875504}} +{"family": "companion", "seed": 504001, "x": [1.3454452487919029, 1.1922288230586457, -0.9711130930480547, -3.194161177589338, -0.37723292929322705, 2.61869869868932, 3.2145255716869676, -5.0837138278546465, -1.1167841449543592, -1.8837499413215197, 5.906706474268496, 0.40365313533084324, -1.419202166236122, 1.2050451450162998, 2.3904933750071438, 1.3821567296856507, -1.038057552541133, 1.0553134327493976, 0.05027133809735837, 0.09842091228662767], "ratio96": 1.9198559358004614, "ratio_final": 1.9198552462314853, "diag": {"ratio": 1.9198552462314853, "nonnormal": 0.7179767188776328, "eig_gaps": [0.2102421814648462, 0.466291139510505, 0.5492546702574377], "discness": 0.06235399989261289, "peaks": 4, "deg": 3, "diamW": 10.638268180538134}} +{"family": "peaks", "seed": 505001, "x": [0.7162312206223256, -0.5678662675564179, 0.05469255179749438, 0.40709325674373453, 0.8472446263569182, 0.44327986071337255, 2.2825714050836785, -1.2047064402323389, 0.16133187207321162, 0.2166060771203838, 2.554176143583713, 0.7620838388542446, -3.4356218662527964, -0.4414995512905575, 7.503203744328344, 0.38559047016323145, -5.561526836737574, 0.4092303040255224, -0.5888507723656181, -0.21526820440409136], "ratio96": 1.9920052751918345, "ratio_final": 1.9919948550362034, "diag": {"ratio": 1.9919948550362034, "nonnormal": 0.6826737683836362, "eig_gaps": [0.2029986789589923, 0.2608809458804492, 0.3014643791181786], "discness": 0.014751377828202242, "peaks": 4, "deg": 3, "diamW": 3.908290233612499}} +{"family": "real", "seed": 506001, "x": [-0.2565006746329338, -0.2499159796967345, 0.18278277154186506, 0.8909611830041402, 0.22778763762963034, 0.29424936447882677, -0.011571650738213852, 0.16146729510346097, 0.7800526918537881, 0.05282602667020613, 3.8830013400074908, -0.975881773003448, -0.651445956736074, -0.9809025663521542, 1.863644370374139, -0.490594699480808, -0.0028718027702500978, -0.001071210878220749, 0.0024934730725696295, -0.0005913766766596903], "ratio96": 1.9986249594658858, "ratio_final": 1.9986293893348943, "diag": {"ratio": 1.9986293893348943, "nonnormal": 1.3428985634832233, "eig_gaps": [0.14481573915829782, 0.17628848170234968, 0.29585412629343966], "discness": 0.005808180815344595, "peaks": 3, "deg": 3, "diamW": 4.13219295958844}} +{"family": "random", "seed": 500002, "x": [1.079166601601878, -0.5947810552038502, -0.01456172020084387, 0.6732708502772535, 0.05408658682694999, 0.29683281431082253, -1.9216586489470302, -1.2065669928043672, 0.4569821742136546, 0.18507263046558142, 0.8770564186705423, 2.619828477181286, -0.9310665059485426, 0.09905141295637185, 0.8535636164437238, -0.938871811718955, -0.49065554969071146, 1.2092072668239138, -0.04494175098720074, 0.15162035993734113], "ratio96": 1.9823948360410084, "ratio_final": 1.9825954160573804, "diag": {"ratio": 1.9825954160573804, "nonnormal": 0.7184301990431327, "eig_gaps": [0.0954267889141793, 0.3388136705109138, 0.41761595480531694], "discness": 0.044991606393835114, "peaks": 4, "deg": 3, "diamW": 4.009841435655586}} +{"family": "jordan3", "seed": 501002, "x": [0.14126090417176138, -0.2257875737725679, 0.15072544542741897, 0.4887234525794647, -0.21582611246277267, -0.7768366437564411, 1.2178109970394457, 0.26884267596875044, -0.5380223155210195, 0.7968889033477157, 1.2136607390013092, 0.021033706287907485, 0.15643658027490032, -0.22369292829645748, 0.06364145519185843, 0.46519547776621406, 1.7441812814206237, -0.7985980083306673, 0.009155360496421808, -0.3706333013958898], "ratio96": 1.9746908601982593, "ratio_final": 1.9746841950990968, "diag": {"ratio": 1.9746841950990968, "nonnormal": 0.742903002605996, "eig_gaps": [0.27759081101808597, 0.3020854144965906, 0.5570034497313459], "discness": 0.078523427170913, "peaks": 4, "deg": 3, "diamW": 2.3654690819787434}} +{"family": "jordan2", "seed": 502002, "x": [-0.5624697530949847, 0.4541501469672754, -0.3195251898566118, 0.3606021720878767, 0.9847758248860997, 0.15617389101677298, 1.2613524298033836, -0.9400003824672798, 0.019316578113084563, -0.11086746512944498, 0.02004632513240958, 0.09036710814475905, 0.17690553075938986, -1.0447647558288553, 1.1940054661990773, -0.6366325070160492, -0.33990221657859965, 0.35820916522456925, -0.11023382899490108, 0.12635023147027286], "ratio96": 1.8504650973084025, "ratio_final": 1.8437007000903027, "diag": {"ratio": 1.8437007000903027, "nonnormal": 0.8428970532761042, "eig_gaps": [0.11571003668080475, 0.5867991405895749, 0.7003401477798117], "discness": 0.2876617934303926, "peaks": 4, "deg": 3, "diamW": 2.2498741979398282}} +{"family": "nearnormal", "seed": 503002, "x": [-0.9905575749733214, 0.7146753300710389, 0.5608119001140869, 1.2287606375297375, -0.11492334522720735, -0.7164634425604411, -0.22272187601176902, -0.058908899535744054, -0.5615770039077239, 2.06185892988116, 0.0670065399924976, 0.06634000700909656, -1.0620917050512833, -2.3400728879323456, -4.920346419227165, -2.4779264310825573, 2.923001315095547, -0.28494696524923036, 0.7432508387047566, 0.535732399700767], "ratio96": 1.5483427824386224, "ratio_final": 1.5768934243014407, "diag": {"ratio": 1.5768934243014407, "nonnormal": 0.9801960261134606, "eig_gaps": [0.5859229767047134, 0.6014950422817369, 0.7382617410942227], "discness": 0.11000412595186242, "peaks": 3, "deg": 3, "diamW": 2.7893237924236307}} +{"family": "companion", "seed": 504002, "x": [0.13141908225047094, 0.5047987259507355, 0.29658858316490755, -0.41687424220017116, -0.5506997716056026, -1.9835134020835696, 1.5597868784088615, 0.16194955206448536, 5.5647024766702e-06, -1.273270165898631e-05, 1.6863752470423533e-05, 8.654171542564112e-06, -0.020253105774423685, 0.21654849374475144, -0.14667118719202327, 0.35968735268161456, 0.29554001836849797, -0.07799626145338417, 1.1031540098269728, -0.02311857945943209], "ratio96": 0.9999999999739742, "ratio_final": 0.9999999999848634, "diag": {"ratio": 0.9999999999848634, "nonnormal": 0.5601991160367072, "eig_gaps": [0.30693788359577445, 0.5838398438471112, 0.8457623830771873], "discness": 0.4109165248449733, "peaks": 1, "deg": 3, "diamW": 3.0506358603388426}} +{"family": "peaks", "seed": 505002, "x": [-1.2473312389132412, 0.03483507952341347, 0.5960233822065992, -0.022482513288902994, 0.6411452621046123, -0.9466044008936096, 3.1029711060125162, -0.47929223035202656, -0.2865973169105285, 0.6008059211668431, 3.5236827012701433, 0.10375408036711858, -1.8921514392618834, -1.203957732127999, -1.1613934911893078, 1.7721158091576523, 2.438359456192721, 3.322899959800054, 0.0746165113870571, -0.24761317777520375], "ratio96": 1.9813651608710636, "ratio_final": 1.981347870661632, "diag": {"ratio": 1.981347870661632, "nonnormal": 0.7110121093359228, "eig_gaps": [0.17558135410788492, 0.3499861046472854, 0.40388749292477644], "discness": 0.03605227599254915, "peaks": 3, "deg": 3, "diamW": 5.269482139423678}} +{"family": "real", "seed": 506002, "x": [-1.9744829140393643, -0.8399882907828158, 2.259738070149643, 1.1104163517102132, 2.5011186393930487, -0.15750267038195365, 0.752056089505021, 1.9131115974699584, 1.0227611939658399, -1.3263516109788505, 0.2665817986910771, 0.07320527625896234, 0.26833747542995623, 1.1868505946426744, -13.705605780231839, -0.5019332022684981, 1.10471242329622, 0.2240006052298853, -1.9186357631211934, 0.7505899783125412], "ratio96": 1.2827768318740689, "ratio_final": 1.287282698565298, "diag": {"ratio": 1.287282698565298, "nonnormal": 0.8173469118523187, "eig_gaps": [0.2431218863724856, 0.8527952394211679, 0.8781297236251565], "discness": 0.22252116856972248, "peaks": 2, "deg": 3, "diamW": 5.308822693759193}} +{"family": "random", "seed": 500003, "x": [-0.715308304058238, -0.7231779561539313, -0.13033034729365361, 0.9093040880527814, -0.8861335077932107, -1.2060453053955138, 0.24806495176617993, 0.14425334429071718, 2.9697399364565578, 0.4663671992251304, 1.338263799926328, 2.9527004148812477, -3.030125335789352, -0.9106344249467557, -2.946471786311541, 0.8084999847297709, 0.03516201708643593, -0.15964510868482729, -0.029539834404597815, -0.00752653130309763], "ratio96": 1.9349121215729068, "ratio_final": 1.9359821267482429, "diag": {"ratio": 1.9359821267482429, "nonnormal": 1.2768319404358481, "eig_gaps": [0.10217253315737142, 0.3459243000770401, 0.448096186196038], "discness": 0.06393708645344323, "peaks": 4, "deg": 3, "diamW": 5.013024843525888}} +{"family": "jordan3", "seed": 501003, "x": [0.5577604597064894, 0.26501883573393803, -0.08759005459389191, -0.282740039493834, -0.26140095148288356, -0.3952171977108754, -0.5456429414666236, 1.277780618277661, 0.02202211475810789, 0.04487673316807245, 1.518859221133376, -0.3977283305738564, 0.04906980580102918, -0.15827825384978667, -0.3839636005093281, -0.08795002044246919, 0.9740185719077649, 0.8416296161948017, 0.2438031105306604, -0.0743518230313806], "ratio96": 1.958941064105588, "ratio_final": 1.9594882050402616, "diag": {"ratio": 1.9594882050402616, "nonnormal": 0.7037578899594588, "eig_gaps": [0.08728840115055163, 0.3568921065045617, 0.44359345622539814], "discness": 0.05179686475545965, "peaks": 3, "deg": 3, "diamW": 2.371790518197552}} +{"family": "jordan2", "seed": 502003, "x": [0.14662944646952766, -0.10673327840047382, -0.23098812019344794, -0.4187386068919553, -0.2691805248312558, -1.0577093654039518, 1.4398323586109718, 0.8273259996840779, -0.04598085778348075, 0.14244362953644502, -0.07644354790311306, 0.04139629027584234, -0.0874244929077668, 0.20153109259657187, 1.0736231761632362, 0.34167306849291046, 0.10962787486450135, -0.07202238719092877, -0.006625271898595066, 0.00619996143280397], "ratio96": 1.9716376557634991, "ratio_final": 1.972273657766895, "diag": {"ratio": 1.972273657766895, "nonnormal": 0.9669712478315908, "eig_gaps": [0.2816454424314092, 0.3680481741268153, 0.5967716990950583], "discness": 0.020767633998951637, "peaks": 3, "deg": 3, "diamW": 1.7392048090791261}} +{"family": "nearnormal", "seed": 503003, "x": [-0.5622071713189568, 1.0484692032467515, -0.1258844537520527, -0.787777972177629, -0.5303588755019489, 0.17556588943707369, 0.6171553996370316, 1.391718497682322, 0.9119404244102896, -0.9214545443532596, -2.199831896952622, 0.8786679420236467, 2.1638443276502493, 0.21786843988738036, 1.1440069021816004, -3.6348165084616513, 0.5384246951769596, -0.3947311591286482, 0.06062071609798446, -0.04663732821447272], "ratio96": 1.8524578282532453, "ratio_final": 1.8592570938725468, "diag": {"ratio": 1.8592570938725468, "nonnormal": 1.1519147239720757, "eig_gaps": [0.2585188090777314, 0.3092253239668432, 0.5585925460925428], "discness": 0.046518830997230164, "peaks": 3, "deg": 3, "diamW": 3.378802973247606}} +{"family": "companion", "seed": 504003, "x": [-1.2644211293910106, -1.3016927261452884, 0.3709218147734667, -0.6162515363404695, -1.0786114985851927, -1.7153717828522685, -0.1573649704175326, -0.40518164901645926, -0.53642279689854, 0.5559109827149421, 0.21080502866483053, 0.18923319665354532, -1.2854640129119081, 0.4453704584437431, 0.6454823698499458, -0.02641882628190919, 0.2021726082130687, 0.6424159478123455, 0.5594934163598275, 0.28867723087423935], "ratio96": 1.5764737279162881, "ratio_final": 1.5720460083945795, "diag": {"ratio": 1.5720460083945795, "nonnormal": 0.18232005062994341, "eig_gaps": [0.20363738589590027, 0.7962339652765743, 0.8168636455704699], "discness": 223377024353.84473, "peaks": 1, "deg": 3, "diamW": 2.2269616405524753}} +{"family": "peaks", "seed": 505003, "x": [-1.3173909570640232, -0.04806319439464175, -0.22161130503647225, -0.31929287991635824, 0.3429392048734153, -0.024468210005288475, 1.6052861631331583, 0.2980863017313476, -0.16542978022622318, -0.29254571460665246, 0.012214540847202094, 0.01161458023389096, -2.2306113217603913, -1.4209412021671444, -2.4100672453262035, -0.40289975456735183, 1.2381411773904203, 0.4266890635572962, 0.4564300709791528, 0.22879188838872633], "ratio96": 1.8534265506140137, "ratio_final": 1.8616595183047597, "diag": {"ratio": 1.8616595183047597, "nonnormal": 0.9984655437224198, "eig_gaps": [0.29639294655850634, 0.525331794413327, 0.7727453348111146], "discness": 0.3756073715048909, "peaks": 3, "deg": 3, "diamW": 2.1488292883344355}} +{"family": "real", "seed": 506003, "x": [-1.2080906630373835, -0.7811207897958533, -1.1195841476372272, 1.0345494421196282, 2.6276581436218707, -0.850597606771537, -3.443115191884935, 0.7570862126481509, 0.959278671554517, 1.7839468803866219, -2.3346581999626213, -0.47946775551561716, -4.2014664811727105, 2.169204016777977, -1.0972488931203304, 1.0788164767425936, 1.8485943836818886, 0.5517116761443153, 0.07764830133492373, -0.07342991020672082], "ratio96": 1.8642393755887576, "ratio_final": 1.8660990242562745, "diag": {"ratio": 1.8660990242562745, "nonnormal": 0.7317393819973328, "eig_gaps": [0.2782673622561853, 0.5872612172160051, 0.6421138306027048], "discness": 0.11494911667945612, "peaks": 3, "deg": 3, "diamW": 6.532660183356133}} +{"family": "random", "seed": 500004, "x": [0.09443646283812443, 1.765449821709887, 1.2320396193054757, -0.18546029232021768, 2.903672737728493, 2.1854671867621223, 0.11541977117164072, -0.09177120863394156, 2.3271303780865797, -0.2644608679375053, -2.9303502566095196, 1.6644332309452257, -1.4087538164228475, -4.519675698247865, 0.4777348511164487, 0.43174499570285146, 0.5894737798837215, -0.0267366511770797, -0.025787958336093042, 0.039865754808739426], "ratio96": 1.7181772332567675, "ratio_final": 1.7182643773409472, "diag": {"ratio": 1.7182643773409472, "nonnormal": 0.8294132665945488, "eig_gaps": [0.4552150302725093, 0.5725485396982, 0.5847460673880731], "discness": 0.3983510821991655, "peaks": 3, "deg": 3, "diamW": 4.961084637317503}} +{"family": "jordan3", "seed": 501004, "x": [-0.1746341831548326, 0.3145458174334239, 0.10659994484091848, -0.1163687649503421, -0.19041755665805288, -0.28126637967842144, 0.9470418251354056, -0.3346179619680207, 0.08270956933747875, 0.037129517538420406, 0.8739519035789782, -0.5988611587734088, 0.058549100054516044, -0.0006001775963409082, 0.18056774791411653, 0.07373792650645855, 0.9388517570738802, 0.05126126457356961, -0.2093053104656878, -0.13385437254405727], "ratio96": 1.9690107251209321, "ratio_final": 1.971277939419618, "diag": {"ratio": 1.971277939419618, "nonnormal": 0.7547627528563305, "eig_gaps": [0.21472318236310392, 0.3252360274008694, 0.3767192107898742], "discness": 0.02466560465157238, "peaks": 3, "deg": 3, "diamW": 1.5821365044431808}} +{"family": "jordan2", "seed": 502004, "x": [0.029770918540695426, 0.013557312220732073, -0.15977660775375163, -0.33363837757640225, 0.1098118399489057, 0.0767037826395883, 0.644499200718647, 0.4853901970700765, 0.3331325909300783, 0.5431415590401549, -0.0665658024729473, -0.466338091303627, 0.041145155195050576, 0.12457357806795005, 1.0924514807698684, -0.0017601876873711367, 0.05324737309374375, -0.1621131565020537, 0.005834214919578669, 0.04039951879206363], "ratio96": 1.9718541478748715, "ratio_final": 1.972615867427874, "diag": {"ratio": 1.972615867427874, "nonnormal": 1.3189958465037148, "eig_gaps": [0.08587348071681365, 0.333186219219635, 0.4135504553806763], "discness": 0.024966320158555027, "peaks": 3, "deg": 3, "diamW": 1.1872244424756242}} +{"family": "nearnormal", "seed": 503004, "x": [0.06095555551291205, -0.083736792978448, 0.9111889820095083, 0.33476389729345113, 0.9712318531766985, 0.46378077163990006, -0.000899235563300465, 0.0007690682897037891, -0.00033244121294242254, 0.00039600134347156054, -8.525181562321008e-05, -0.3690307400006697, 0.13086975159764064, -0.23143688097129234, 0.4045386371093611, -0.049515347313990765, -1.0978060106223673, -0.2770076888831402, 0.9856723264127631, 0.2778082087419983], "ratio96": 1.764633951521981, "ratio_final": 1.7991307931216087, "diag": {"ratio": 1.7991307931216087, "nonnormal": 0.09184018604464274, "eig_gaps": [0.11884779552100579, 0.7914455149893928, 0.8871573168118372], "discness": 148900207637.48236, "peaks": 4, "deg": 3, "diamW": 1.1973655190791341}} +{"family": "companion", "seed": 504004, "x": [-1.365896881162123, -0.9432807423943181, -0.8452534096982203, 0.2563704112078332, 1.2659592152279493, -0.8762703518481754, 1.0612093126535482, 1.6302626157109519, 0.574227648050369, 0.9456953263625429, 1.6996663293408303, -1.7246278804079278, -1.0871843632180007, 3.0063428186073526, 5.667182605568151, -0.19741049956292933, 2.5384482576257392, -4.357006186738648, -0.575322904734467, 0.19208072731123826], "ratio96": 1.848448954940936, "ratio_final": 1.8727094280973273, "diag": {"ratio": 1.8727094280973273, "nonnormal": 0.7109458838730016, "eig_gaps": [0.3069856859548061, 0.5624061261852886, 0.6180070245209218], "discness": 0.0954617661388771, "peaks": 3, "deg": 3, "diamW": 4.259998573538774}} +{"family": "peaks", "seed": 505004, "x": [0.27815110116216446, 0.26424729329956953, 0.49918653906930366, 0.23745530448622304, 1.3092797513235106, 0.08769151601423145, 0.099587768566181, -0.12892805755074208, 0.368868495019229, 0.10484222845517964, 0.8475634895113329, 0.8218501699120739, 2.786182809227646, 1.1593813720719077, -2.5159470710738248, 0.17098400000631025, -1.3419344062036527, -1.2513353950627608, 0.5570338213136496, 0.3848045969635858], "ratio96": 1.940438693852559, "ratio_final": 1.9439309034314967, "diag": {"ratio": 1.9439309034314967, "nonnormal": 0.6934884741200089, "eig_gaps": [0.15023467118501335, 0.5558705913065035, 0.7058766224959847], "discness": 40720464314.08041, "peaks": 4, "deg": 3, "diamW": 1.482036477554378}} +{"family": "real", "seed": 506004, "x": [-0.19296662767299722, -1.1466083489249699, 1.6684144427296443, 1.1515756335479441, 2.6298097261134785, -0.20413891702813436, 1.7375960387690688, 0.3338091567189336, 0.5028708787442547, -0.0779469676873566, -2.177656808763828, 0.22206767952174902, -0.4365914809707139, -2.772007601634985, -2.886306284173396, 2.734238081708968, 1.2854400520018454, -1.4466176540843456, 0.04311411472130805, 0.3002295317268304], "ratio96": 1.8437084745447607, "ratio_final": 1.8437241614229833, "diag": {"ratio": 1.8437241614229833, "nonnormal": 0.5225718939628375, "eig_gaps": [0.3820069926373305, 0.6797587115936086, 0.6840170258638105], "discness": 0.3537683706612981, "peaks": 4, "deg": 3, "diamW": 4.350704780585473}} +{"family": "random", "seed": 500005, "x": [-2.069087991696604, 1.846251733041055, -0.6461220354212477, 1.3069586145382845, 1.8690967294694711, 0.061165639939162894, 0.2999667520400482, 0.30004074115845636, 2.7346458905501414e-07, -0.2300046831896692, -1.7280800613736385, 1.1311797804746404, -2.3090807425558726, -0.39123693455312414, 1.922363775231074, -0.2737550967542935, 0.32342386830418157, -1.3695032017575492, -0.7046835200210926, -1.6594576786847801], "ratio96": 0.9999999999695706, "ratio_final": 1.0229491721467197, "diag": {"ratio": 1.0229491721467197, "nonnormal": 0.5757398622173365, "eig_gaps": [0.32369047129240713, 0.5970471356753759, 0.9197378236866923], "discness": 0.5493964662242448, "peaks": 1, "deg": 3, "diamW": 4.701195836329327}} +{"family": "jordan3", "seed": 501005, "x": [0.4570628799091081, -0.44529633184941947, -0.0608843119631914, -0.41388497600736174, -0.5981879725053629, -0.35143321185530885, 1.4724286133004298, 0.3460775045853062, 0.11899381183506011, 0.08711187384368109, 1.3086284792302691, 0.425579182398219, -0.27846828497663934, 0.11378403269398368, 0.24281562826250874, 0.7973600507965855, 0.9425746239490878, -0.13947498728706625, -0.11503539414762576, 0.018230992363778453], "ratio96": 1.9910387394827864, "ratio_final": 1.991434563410379, "diag": {"ratio": 1.991434563410379, "nonnormal": 0.6510105005408897, "eig_gaps": [0.22547219616273434, 0.23504126099530165, 0.4603385189360222], "discness": 0.06555872874232109, "peaks": 4, "deg": 3, "diamW": 2.3013870931919884}} +{"family": "jordan2", "seed": 502005, "x": [0.15919421226264957, 0.04347563590027815, -0.1971257888099483, 0.0740641677200744, 0.19888345549799358, 0.09251555704917194, 0.6445481008248621, 0.4631113163716093, 0.2022965820124849, -0.9049485275262567, 0.09983030240558963, 0.5029770490281956, -0.06050894556624314, -0.1800735256268043, 2.5620678460187447, -1.0683125867008954, 0.045336270215217876, -0.16527779098743664, -0.11347188151143842, -0.015109274823685822], "ratio96": 1.9900613248531611, "ratio_final": 1.9900582366896558, "diag": {"ratio": 1.9900582366896558, "nonnormal": 1.3625083941209595, "eig_gaps": [0.0460751398593191, 0.2611869919164508, 0.2895297360875586], "discness": 0.014011156991762483, "peaks": 3, "deg": 3, "diamW": 1.369250953130337}} +{"family": "nearnormal", "seed": 503005, "x": [-0.7592780891492801, 0.961520811570451, 1.0578117427317606, 0.9836206167029982, -1.991411636052115, -1.254921000573472, 0.9164382545946852, 0.04378724456273925, -0.7945787582888213, 1.4003280646601892, 0.3205448694452452, -0.8275076854905372, -0.18077706391441456, -1.1202837296346055, 0.5015972216151467, -0.6362871644273103, 1.938377974268537, -0.9515683811755578, 0.27345799424469786, -0.26604744593393936], "ratio96": 1.4123978846974081, "ratio_final": 1.5129410425167429, "diag": {"ratio": 1.5129410425167429, "nonnormal": 0.6216950295816026, "eig_gaps": [0.426300455559366, 0.5948927565566945, 0.8873794556545801], "discness": 0.38185934327894167, "peaks": 3, "deg": 3, "diamW": 4.262778034386288}} +{"family": "companion", "seed": 504005, "x": [1.5814236674850601, -0.8721879896223544, -1.0972020823466826, -0.46230097708718043, -0.11403948352747335, -0.9216309586478708, -0.44298277326437796, -2.0076884067079925, -0.24945458909342266, -0.7165680683500626, 1.9195274011876828, 0.8065171374852447, 3.049156693192973, 1.5951112509697427, 1.0010514159772717, -1.8422829357060242, -0.9365331942570867, -0.03465594031179179, 0.6223160897966367, -0.2751499163917813], "ratio96": 1.6185508319284885, "ratio_final": 1.6276224592666562, "diag": {"ratio": 1.6276224592666562, "nonnormal": 0.6703509043694417, "eig_gaps": [0.2662599682714382, 0.4161800529859599, 0.6648847349590461], "discness": 0.16449730865120551, "peaks": 3, "deg": 3, "diamW": 4.075601201234808}} +{"family": "peaks", "seed": 505005, "x": [-0.7612558511373327, 0.9485626067958162, -0.2428870975528662, -0.6495226125549602, 0.5943327224549597, 0.8616030889176765, 1.8337344297505629, 0.6481842998416011, 0.1485285994921452, -0.2137414249253975, 1.7682802179326647, -0.321043210951741, -1.6746131966458149, 1.550421096824149, -3.3759773378753573, -3.0401967764682727, 1.191152161069493, -3.2718891288469067, -1.0170901910598675, -0.6620272776109222], "ratio96": 1.9117985491108143, "ratio_final": 1.9123656411555223, "diag": {"ratio": 1.9123656411555223, "nonnormal": 0.7457271605630227, "eig_gaps": [0.4319571663780295, 0.5342497945927659, 0.5493540280096324], "discness": 0.03086325421633432, "peaks": 4, "deg": 3, "diamW": 3.1446981195378285}} +{"family": "real", "seed": 506005, "x": [-0.5340120983283243, 0.4255701872537909, -1.3075161397115775, 0.2103137030505705, -0.13958026211815006, 0.133790541829984, 1.1988638191140861, -0.03434047700854527, -0.5240926456713064, 0.4274002669151292, 0.754718722250387, -0.1511003311793701, -1.3047854871209248, 0.4531477395117032, -0.9769323102494223, 0.08457192779511816, 0.5605575174556912, 0.07816684378213559, 0.13138651462636897, 0.038345216910724775], "ratio96": 1.9133284137402589, "ratio_final": 1.913455138646116, "diag": {"ratio": 1.913455138646116, "nonnormal": 0.8287663979654507, "eig_gaps": [0.2617298706796341, 0.42831619207971355, 0.6243868649767325], "discness": 50903268374.86762, "peaks": 4, "deg": 3, "diamW": 1.874543103645356}} +{"family": "random", "seed": 500006, "x": [-0.41204351132940115, -0.23741187869780533, 0.592671956696535, 0.5705249481514083, 0.3769118752944296, -0.04123990295187596, -0.6385099760625764, -0.0062821526399785286, 0.7228527309761854, -0.04738617696154368, 0.0836510598893102, 0.861971928623388, -0.6933115250375579, 0.3837853710985154, -0.43070744085808893, 1.0851426872305678, 0.32152530579968874, -3.2260321702965324, -1.1647415666738032, 1.0639066177465153], "ratio96": 1.7705055706620754, "ratio_final": 1.8827341824610222, "diag": {"ratio": 1.8827341824610222, "nonnormal": 0.7501048303278071, "eig_gaps": [0.35460385985594883, 0.4444062862065038, 0.7047656710825362], "discness": 0.16244895109505675, "peaks": 4, "deg": 3, "diamW": 1.8293586929726868}} +{"family": "jordan3", "seed": 501006, "x": [-0.04099317945531747, -0.1079389890056528, 0.20226388418521069, -0.6741873204955526, -0.01435239544943711, 0.04609913872744986, 0.834863405984223, -0.07857395257133193, 0.5983213951230593, 0.18964257571410637, 0.5194033912620527, -0.28253441969230153, 0.26747345793938815, 1.0946541047294271, 2.113760511701488, -1.5044692306064833, 0.5337996308482178, 0.4312937095305877, 0.2530754550923772, -0.39923175784314635], "ratio96": 1.9545744982898716, "ratio_final": 1.9559317390935447, "diag": {"ratio": 1.9559317390935447, "nonnormal": 1.1725234387015397, "eig_gaps": [0.11331276250277836, 0.4467190977877997, 0.5452017624020895], "discness": 0.12088611434573038, "peaks": 3, "deg": 3, "diamW": 1.3795878092625207}} +{"family": "jordan2", "seed": 502006, "x": [-0.16770548867426338, -0.09385848358388077, 0.1433584941105634, 0.17909134342675526, -0.07200854615851088, 0.2557086653098921, 1.5420857331971387, 0.18378922458726507, -0.5756652218319607, -0.7021691789459082, -0.0034724147345558867, -0.15914449691567167, -0.1351298912452304, -0.24858173300943542, 2.605135825429908, -0.37075718537111907, 0.16033447389032324, -0.09334961144406909, 0.01671396835935704, 0.02277950442062541], "ratio96": 1.9876389785513804, "ratio_final": 1.9878657349927435, "diag": {"ratio": 1.9878657349927435, "nonnormal": 1.3836096000718354, "eig_gaps": [0.12313718823144582, 0.1952344034733779, 0.22292763615400507], "discness": 0.0087149157093932, "peaks": 3, "deg": 3, "diamW": 1.8563809963014617}} +{"family": "nearnormal", "seed": 503006, "x": [-1.015929884400523, -0.4058456269200636, -1.1616772546706022, -0.3693554686800117, 0.15633061522902567, 0.19510581243919597, 0.3740584594215309, -0.48907632687741853, -0.012183374456353743, -0.0019162642343549473, -0.002967672357479615, -0.01239131186041461, -0.5835036135418304, 0.7799393213503123, 1.8937886546641065, -0.38636111535848155, 2.2997654186873944, -0.33052735239152076, 1.0956249129654576, 0.5964199725645171], "ratio96": 1.7655613360193627, "ratio_final": 1.603390187982936, "diag": {"ratio": 1.603390187982936, "nonnormal": 0.17682404839141916, "eig_gaps": [0.08899935189541358, 0.7803262100991092, 0.8493184988563877], "discness": 199309546080.66168, "peaks": 1, "deg": 3, "diamW": 1.6881683655365383}} +{"family": "companion", "seed": 504006, "x": [-0.40707152080667064, -0.3323352698645867, 0.061761814900964135, -0.2859200946784225, -0.6989047737230032, -0.15138172324661517, 0.7115108309266689, 0.2691714017656131, 0.1178459243263385, -0.19627602161907687, 0.7409243915430457, 0.09316032883771098, 0.23852516001208546, -0.2395022537292576, -0.18732685495204637, -0.8501833797750606, -0.1517098589834508, -0.07159919596960171, 1.2676792022126369, 0.3638652184281414], "ratio96": 1.8188771907929755, "ratio_final": 1.8219698678474072, "diag": {"ratio": 1.8219698678474072, "nonnormal": 0.6531346143408411, "eig_gaps": [0.26012287991754646, 0.3568930923932111, 0.5851738321588176], "discness": 0.16717708696567973, "peaks": 4, "deg": 3, "diamW": 1.32007407241262}} +{"family": "peaks", "seed": 505006, "x": [-0.6256433503264893, 0.7712813364613571, -0.38785147574244483, 0.4213983942233402, 0.15874856559299771, 0.2270637541056873, -0.017552209990751758, 0.8595389189655733, 0.5312314995011365, -1.6787254496472575, -0.3450772730947873, 0.5400167769849096, 1.578791288436448, 1.298092735708945, -0.024056010274912547, 4.3493681641162745, -0.18975044869812174, -0.5214739881650416, 0.18959880843139426, -0.07810949504799372], "ratio96": 1.9405249948072203, "ratio_final": 1.9423007722174221, "diag": {"ratio": 1.9423007722174221, "nonnormal": 1.1315153197544892, "eig_gaps": [0.18730863508019316, 0.25685787388847303, 0.4227084985817874], "discness": 0.05626686158109703, "peaks": 3, "deg": 3, "diamW": 2.2585197143533873}} +{"family": "real", "seed": 506006, "x": [-1.2918488970062563, -0.10438299700344839, -0.47265524471854303, 0.015220156592217586, 1.5700104483404893, 0.09490618797059618, -0.9908854020488214, 0.16416240919722266, 0.2900271651558012, 0.29000000279660254, 0.24999895450772702, 0.26000191663941574, -0.39308795296612287, 0.10992210825524883, -1.148670624486136, -0.246928489559761, 0.19784913535130422, 0.13411384787682165, -1.0549001590614566, -0.10930324545228666], "ratio96": 0.9999999999680416, "ratio_final": 1.107756656457773, "diag": {"ratio": 1.107756656457773, "nonnormal": 0.5156420173666064, "eig_gaps": [0.26255102564919564, 0.6482977493695948, 0.9097995945232918], "discness": 0.3248587741345354, "peaks": 2, "deg": 3, "diamW": 3.153210715642043}} +{"family": "random", "seed": 500007, "x": [0.5982539367085842, 0.28087590533948237, -0.2906140044868054, 1.5287685463985716, 0.35830386617967536, 1.3045691643083366, -1.7640176945609476, -0.5295464883022528, 0.9865775741125192, 1.197752756906958, 0.3362129228298452, -2.795257184889161, -0.1467066315824433, -4.601160240095304, 1.2729368957596483, 0.5106827039219279, 0.514489719657443, -0.7490674780614797, 0.14212302564008333, 0.004892357881707492], "ratio96": 1.677985221813004, "ratio_final": 1.6783193414620063, "diag": {"ratio": 1.6783193414620063, "nonnormal": 0.8823679468612136, "eig_gaps": [0.17438630343903028, 0.26706697661712436, 0.3891548754457549], "discness": 0.25692324662435945, "peaks": 3, "deg": 3, "diamW": 3.936986112935207}} +{"family": "jordan3", "seed": 501007, "x": [0.7182254905448828, 0.16030134838626608, 0.9122742906885475, 0.2823604819306559, 1.0624700291923084, 0.37259406962090824, 0.16406860333724005, 0.3820887198250327, -0.48569723101919426, 0.5877173335259569, 0.22899519983624353, -0.19258469839446749, -1.3989498339000828, -0.32940194759614383, 0.8227203132910446, 0.06740023071648579, 1.185550498161223, -0.4264692450052471, -0.3448310032757447, 0.30944420960369423], "ratio96": 1.9607612844343567, "ratio_final": 1.9609974357998128, "diag": {"ratio": 1.9609974357998128, "nonnormal": 0.34689406879099965, "eig_gaps": [0.1767359254835867, 0.23123301778026764, 0.40794791189414875], "discness": 15618905373.987396, "peaks": 4, "deg": 3, "diamW": 0.9914034677881135}} +{"family": "jordan2", "seed": 502007, "x": [0.26080568341964794, -0.2491262742854301, 0.30008698970607117, -0.4355062912243324, 0.24838728265124332, -0.4496276832745092, 0.26041543429098024, 0.06565815412669812, -0.1675685667862731, 0.5650797977444564, 0.044818042902740274, 0.07930371125220992, -0.668099895448035, 0.3882987212505953, 1.3981420171935612, 0.4203852264196202, -0.3321939690202995, 0.5506009384546662, 0.34489907491191385, 0.14488969860468676], "ratio96": 1.986723249553613, "ratio_final": 1.98673415332121, "diag": {"ratio": 1.98673415332121, "nonnormal": 0.5781103498401133, "eig_gaps": [0.07788017843962519, 0.27679030219565603, 0.29191935740306485], "discness": 5698634131.2727165, "peaks": 3, "deg": 3, "diamW": 0.6881544964944972}} +{"family": "nearnormal", "seed": 503007, "x": [-0.31799881471235397, 1.2654732962048028, -1.4693660359741814, 0.4178021774339271, -0.4063411776823115, 0.5368709136224183, 0.37998015139571395, 0.22001915593250287, 0.8379347427050269, -0.42587196533156857, 0.25999597367673916, 0.14998582107063693, -0.2443141686586502, -0.26419516019293754, 0.0772714960614027, -2.730030057576001, 1.6340257235466662, -2.564719508731673, 0.8387196304338773, 0.18069352132027516], "ratio96": 0.999999999971606, "ratio_final": 1.311502965762841, "diag": {"ratio": 1.311502965762841, "nonnormal": 0.35480550185754545, "eig_gaps": [0.40698431166340937, 0.5931558081843561, 0.7928278410905608], "discness": 116307873369.65643, "peaks": 2, "deg": 3, "diamW": 1.8033583498409518}} +{"family": "companion", "seed": 504007, "x": [1.2529431646827507, -1.6422366089660185, 1.3171001441043773, -0.7645633521037738, -1.1030415833394314, 0.43659455750908294, 1.532872301315984, -1.0565037872392107, -1.250121774494891, -0.021375109633892803, 1.8992767229540413, 1.679916347704791, -0.18219845606579294, -0.08278840727167826, -0.32964545327125966, 2.1821154869598542, 0.8146976460219377, -0.20727144490442367, 0.2375512684639946, -0.05634312301504897], "ratio96": 1.8547129945369247, "ratio_final": 1.8730875103984932, "diag": {"ratio": 1.8730875103984932, "nonnormal": 0.7296587493399895, "eig_gaps": [0.18889282047723752, 0.5799396629357269, 0.6744230432901922], "discness": 0.14664118205353546, "peaks": 4, "deg": 3, "diamW": 4.658806162719001}} +{"family": "peaks", "seed": 505007, "x": [-1.2706054051772044, -0.7507613150189849, 0.7803735041622817, 0.1664112393451057, 0.8932209831666835, 0.7370352125749706, 0.938765421948411, -0.039216441478698315, -0.39941338276714944, -0.10461853194911486, 1.68043286769929, -0.001469699185500082, -2.1603587608595274, -2.003173835141376, 0.699606590327484, 0.20822967266532855, 2.9062428915249185, -0.8878581035449749, -0.19968380806051345, 0.4448105416265147], "ratio96": 1.8582262016006263, "ratio_final": 1.8600022210979095, "diag": {"ratio": 1.8600022210979095, "nonnormal": 0.6062565026936417, "eig_gaps": [0.16890257717908858, 0.6523838562616526, 0.7625076883412438], "discness": 0.21605775369531335, "peaks": 5, "deg": 3, "diamW": 3.443851543881442}} +{"family": "real", "seed": 506007, "x": [-0.7372805674436755, -0.1848465849047567, 0.7924225063417616, -0.15553906753021496, -1.4563507622507017, -0.048337109019312416, -4.009466721859688, -0.6036314111396238, -1.861748219024578, -2.1004308114021457, 1.4157142137767416, 0.7291437045249425, -0.3835899512498022, 0.29435297383769843, 1.9518797489724768, 0.9262601437816407, 0.03646633522872182, 0.002670816322623216, -0.007439743793897023, -0.0017536729146357278], "ratio96": 1.9894433643156342, "ratio_final": 1.9896822883667182, "diag": {"ratio": 1.9896822883667182, "nonnormal": 1.3124480003617534, "eig_gaps": [0.1342476160960317, 0.280629870399024, 0.4129387631330127], "discness": 0.020500337106828564, "peaks": 3, "deg": 3, "diamW": 5.451963458496321}} +{"family": "random", "seed": 500008, "x": [-0.03457113233652699, 0.16945064901653617, -0.47024851474979135, 0.38306110816996564, -0.7223470818466753, 0.5933421152358827, 0.24831916881218624, 0.7631290120039853, -0.11653556870476323, 0.10915105826545674, 0.46231100818104187, 0.4681024148912718, 0.049292866210784665, -0.648468430538026, 1.6556312856198325, -1.2072988757047924, 0.37537438961766123, 0.5278892364870751, -1.2759196990172863, -0.6157660690768139], "ratio96": 1.9680422644981976, "ratio_final": 1.9684583081604405, "diag": {"ratio": 1.9684583081604405, "nonnormal": 0.4475722833311422, "eig_gaps": [0.24941425495278938, 0.3686487211061803, 0.6138073650890158], "discness": 0.5486714087413636, "peaks": 5, "deg": 3, "diamW": 1.316228455665601}} +{"family": "jordan3", "seed": 501008, "x": [-0.22967552929408996, -0.6553382113902071, 0.09591273175143344, -0.21016545132886433, -0.447378708905801, 0.23592161109636045, 1.4754058355895234, -0.1659840238350465, -0.12602318140767338, -0.13202976062411786, 1.4817472393081825, -0.04360133906567049, 0.15630587191182743, 0.16240293612136097, 0.6677018847472302, 0.32273319396713585, 1.08417700664884, -0.3486996735121546, -0.13055511072750237, 0.11115942711080645], "ratio96": 1.9778554304637277, "ratio_final": 1.9779558369327876, "diag": {"ratio": 1.9779558369327876, "nonnormal": 0.6907212665393927, "eig_gaps": [0.23980163765553178, 0.3056438368148927, 0.39890624417229414], "discness": 0.03030247786563593, "peaks": 3, "deg": 3, "diamW": 2.299946981333288}} +{"family": "jordan2", "seed": 502008, "x": [0.41367615487311693, -0.7052270593640513, 0.08829083198288962, -0.10135507223690005, 0.34829972436371504, -0.37789811148204056, 0.9632522689023822, 0.2764422942706417, -0.3117553644247148, 0.2191684404669858, -0.002824586931575168, -0.12219485459342572, -0.11564460031266272, 0.5066266995514819, 0.9340751361273734, -0.3449877309284852, 0.27769916620764024, 0.24304319334424207, 0.06224397839123205, -0.2736128288442259], "ratio96": 1.9530161222853863, "ratio_final": 1.9540779197750053, "diag": {"ratio": 1.9540779197750053, "nonnormal": 0.9208109449412336, "eig_gaps": [0.26102842657445924, 0.29683313351057794, 0.536421763741768], "discness": 0.3103013675079361, "peaks": 4, "deg": 3, "diamW": 1.2787642696599515}} +{"family": "nearnormal", "seed": 503008, "x": [-0.03363069120592657, 0.8636346513180722, -0.0938011905603858, 1.1289404598312123, 0.07073659051785171, 0.6741977660886719, 0.4453939021242344, 0.5412399143123099, 0.09157780176287447, -0.09929145938264042, 0.10527323115379554, 0.048135467754098854, 0.3140269723016823, -0.854409041922854, -0.4943593871395099, 0.49222076268449766, -0.28276890888993655, -1.6652245354557493, 0.40719749894289303, 0.06884434114936022], "ratio96": 1.9353014219635578, "ratio_final": 1.9353272570811622, "diag": {"ratio": 1.9353272570811622, "nonnormal": 0.2613718356286785, "eig_gaps": [0.2773329546176574, 0.3488309310596085, 0.6200946181990649], "discness": 10073639129.064056, "peaks": 3, "deg": 3, "diamW": 0.779872043528181}} +{"family": "companion", "seed": 504008, "x": [1.387211197120965, 0.806934086592578, 1.5698213100156604, 1.0240601348635172, 1.419219489927798, 0.8523088880075473, 0.04814582394035173, 0.2513965692415583, 0.2707237505686544, 0.13060816550858242, 0.5374980755144669, 0.12635263712353365, -0.4531184083505695, -7.370400403896171, 3.180732231244413, 0.21492749369755793, 0.15289407294122048, 0.29592472855422336, 0.5538826064788522, 0.5287440988515064], "ratio96": 1.9125944932253118, "ratio_final": 1.9146148700527859, "diag": {"ratio": 1.9146148700527859, "nonnormal": 0.07064830910752298, "eig_gaps": [0.07887069942619285, 0.3244517001282228, 0.4029692912045008], "discness": 5489290122.492132, "peaks": 3, "deg": 3, "diamW": 0.7040434819590969}} +{"family": "peaks", "seed": 505008, "x": [0.5596459828032446, 0.44411317547688045, 0.5533300193677884, 0.0169462314332669, 1.2114364357569096, 0.019783570030853626, 0.010562561074090112, 0.08394017046089772, -0.03613876029336126, 0.2015246018398989, 1.0809734299510376, -0.7137323306054711, 3.0389462166199923, -0.4117799292534132, -2.798685527084828, 0.41308682599982816, -1.0742999886223332, 0.25874105128285885, 0.4152775474949746, -0.09139246136532], "ratio96": 1.9800725476913412, "ratio_final": 1.9802751031818326, "diag": {"ratio": 1.9802751031818326, "nonnormal": 0.6776918652028181, "eig_gaps": [0.2911664210120919, 0.4485350076101817, 0.5300698427835712], "discness": 26694510731.97992, "peaks": 4, "deg": 3, "diamW": 1.4672489807915494}} +{"family": "real", "seed": 506008, "x": [-0.28458217987588463, 0.09780222291061479, -0.7126645796094393, 0.2413941537692197, 0.42375745146955834, -0.04810111770192084, -1.1838661829171668, 0.07822944898160443, 0.6755570565222454, 0.1543722884672247, -1.148010560031732, 0.051340400228365925, -0.10880549570683819, 0.01968204432621616, 0.3304377392629603, -0.1416025141469725, 0.8057197129020177, 0.07980509790805379, -0.06616076915753935, -0.07685133089051241], "ratio96": 1.9712091093854216, "ratio_final": 1.973556833656473, "diag": {"ratio": 1.973556833656473, "nonnormal": 0.6998178361065162, "eig_gaps": [0.21136005402167093, 0.3385379257237883, 0.5489536367057761], "discness": 0.07329277387037898, "peaks": 4, "deg": 3, "diamW": 2.13627491600623}} +{"family": "random", "seed": 500009, "x": [0.17283796570937904, 0.000932560475767798, 4.094841522668904, 2.0984402361351027, -0.9334244965510919, -0.28880150583732256, 2.531047066902074e-07, 9.033371110550715e-07, 0.7805629224037461, -1.4103742411905227, -3.006570006865509e-07, 2.6345838054169797e-06, 0.759265398313272, 0.9139806879430488, -0.6246164842289371, 0.33649451374968065, 0.46010058333015974, -0.052576551724925105, 0.7002803242832355, 1.0455120650446306], "ratio96": 0.9999999999724575, "ratio_final": 0.9999999999999187, "diag": {"ratio": 0.9999999999999187, "nonnormal": 0.18201175734948313, "eig_gaps": [0.19100856090928922, 0.7428816982398109, 0.9297063617270538], "discness": 1.3998786183233334, "peaks": 10, "deg": 3, "diamW": 5.987032304476571}} +{"family": "jordan3", "seed": 501009, "x": [0.2821294326834504, 0.36895800863397477, -0.5650967479501139, -0.4809262717654036, -0.23654366878070246, -0.3016783921359869, 1.5234350749463461, -0.09508698764993168, -0.01850288102992754, -0.2355500170340836, 1.679901544510132, -0.5533669827441372, 0.11881582558146447, -0.1247943763454511, 0.20713268358299985, 0.36770117337767627, 1.19178189612852, 0.42610651674449096, 0.18633425420667798, 0.010331430572217908], "ratio96": 1.9853044653358427, "ratio_final": 1.9852625351617286, "diag": {"ratio": 1.9852625351617286, "nonnormal": 0.7188103264564445, "eig_gaps": [0.14166066587090936, 0.320894627625591, 0.4542151540130005], "discness": 0.05541818691616323, "peaks": 4, "deg": 3, "diamW": 2.6420073763429883}} +{"family": "jordan2", "seed": 502009, "x": [0.026712999691534373, -0.017608943440771047, 0.04533415958651206, 0.39481666014206296, 0.021963244789375502, -0.09676162876056095, 0.8263693109113025, 0.2581495327098104, 0.2258348478589608, -0.22274932996377733, 0.23979487022154578, 0.13891871288793012, -0.02316074091573289, -0.21040667267026086, 1.1209809286783665, 0.19513514847400099, 0.09865750619618785, 0.04023938231473487, 0.12895462542856467, 0.06926039638386904], "ratio96": 1.9724903402031084, "ratio_final": 1.9773480551859004, "diag": {"ratio": 1.9773480551859004, "nonnormal": 1.271655869474597, "eig_gaps": [0.07693685862047848, 0.4005678689049317, 0.4774976438989443], "discness": 0.041011149981340894, "peaks": 4, "deg": 3, "diamW": 1.0306512267767316}} +{"family": "nearnormal", "seed": 503009, "x": [-1.240817047639691, 1.7307599538166303, -1.991417356877642, 0.30037302785604053, 1.0987906982001805, 0.789919369628617, 0.2799974671855091, 0.2800100456146745, -0.38649321769613315, -0.4065612671325171, -0.02000845039108219, -0.17999054185869096, -2.1433252735345816, 0.4327651996219883, -1.9427293495394085, -4.100122259006257, 1.7107713304483945, -3.3911718221899374, 0.5321150702561429, 0.6156685682881737], "ratio96": 0.9999999999773331, "ratio_final": 1.176328823444026, "diag": {"ratio": 1.176328823444026, "nonnormal": 0.21560830750260926, "eig_gaps": [0.5033110836659874, 0.7857024052558023, 0.9748451109029588], "discness": 1.7041508459188444, "peaks": 2, "deg": 3, "diamW": 3.2094784626218775}} +{"family": "companion", "seed": 504009, "x": [0.5713728664479059, -0.4847888733538984, 0.10566944107922101, -0.6964589370727375, 0.3053996760380849, -0.6218330471296867, 0.6880471570791842, -0.3319955478023392, -0.23386530412370654, -0.15153621120772764, 0.7892353558418967, -0.34227455440219134, -0.8649491859269002, 0.1170281344246965, 0.8569729830503166, 1.5056154621269964, 0.6344866166543877, 0.05380878648622285, 0.9928131466396728, -0.4775055028321879], "ratio96": 1.934878344455013, "ratio_final": 1.9352691803344193, "diag": {"ratio": 1.9352691803344193, "nonnormal": 0.4278927248596357, "eig_gaps": [0.1657354153421815, 0.23257437066057263, 0.39763391476203364], "discness": 22675913627.53026, "peaks": 3, "deg": 3, "diamW": 1.286486045668693}} +{"family": "peaks", "seed": 505009, "x": [0.5614319949003304, -0.2948969282292919, 0.6998432198189439, -0.0984344722748367, 1.2142675173130146, -0.24424003445680253, -0.07937631299600059, 0.10446992353135731, 0.47145925098837377, 0.6092599200426175, 0.5658669475234552, -0.6160005820088137, 1.6944775255460751, -2.524344360043127, -2.0408678839332053, 1.425438903285118, -0.6424777783561826, 1.1432053664846555, 0.343505227658069, -0.3047366146381639], "ratio96": 1.9696616946033516, "ratio_final": 1.9697441665215818, "diag": {"ratio": 1.9697441665215818, "nonnormal": 0.547950833538861, "eig_gaps": [0.18979424057069688, 0.42226815636816234, 0.517124272443979], "discness": 18751508051.343136, "peaks": 4, "deg": 3, "diamW": 1.266229363716848}} +{"family": "real", "seed": 506009, "x": [-1.5032556651800535, 0.11389351538625425, 0.20704136140636978, 0.32239969964942733, -1.7506050192922924, 0.04623322560446694, 0.0007862150973544763, -0.00030156154302410316, 0.9481511715700149, -0.15296831108148917, 3.6438566880762207e-06, -0.0008522075254251533, -1.5041046737727004, 0.1267412717502434, -0.09075669229494204, -0.1425856571697962, -0.3183260692135189, -0.04646083745178978, -0.4352097620856077, -0.000312922089239914], "ratio96": 1.7512202359166855, "ratio_final": 1.7520955385073727, "diag": {"ratio": 1.7520955385073727, "nonnormal": 0.21067588593856093, "eig_gaps": [0.10927789569813624, 0.7342227686392612, 0.8424922672857671], "discness": 265281436732.8345, "peaks": 3, "deg": 3, "diamW": 2.346644579239637}} +{"family": "random", "seed": 500010, "x": [-1.5095708789915676, 0.7394727232674816, 1.3099777971542839, -1.7386103798494057, -1.450378615302026, -1.7322112575725688, -2.873906860112511, -0.6776298407961584, -1.0680489412469576, 0.778068643343907, 1.392028999591648, 1.7234568350734734, 0.43788511827129317, -1.0309509104694448, -1.2282113646994133, 1.0333978880829524, 0.8864883518651041, 0.8895676117093843, -1.0232162024983886, -0.6042582550113238], "ratio96": 1.4465717235015545, "ratio_final": 1.45736747551724, "diag": {"ratio": 1.45736747551724, "nonnormal": 0.746666731857822, "eig_gaps": [0.4550946489282947, 0.5081016589498918, 0.6909575218661724], "discness": 0.11690930315995035, "peaks": 3, "deg": 3, "diamW": 5.432699896017127}} +{"family": "jordan3", "seed": 501010, "x": [-0.1144253048693335, -0.18924399227146765, 0.3437064990649983, 0.5279498071967754, -0.5986328974211959, -0.6152520348467372, 2.069236341559678, -0.9737303289213511, -0.10551299240256229, 0.2872178382139535, 2.040411182727972, -0.23444894698092525, -0.08677404325123467, -0.4081438909213334, 0.7872413684801476, 0.377018112658817, 2.2217083587688857, -0.5781350325434547, -0.04121406317150289, 0.248897187737942], "ratio96": 1.9913879621825537, "ratio_final": 1.9915603286186754, "diag": {"ratio": 1.9915603286186754, "nonnormal": 0.7398884071167791, "eig_gaps": [0.18942661696375726, 0.24995999116958895, 0.43514549689129317], "discness": 0.0357252621042379, "peaks": 4, "deg": 3, "diamW": 3.404664008414236}} +{"family": "jordan2", "seed": 502010, "x": [-0.30389812614964806, -0.23728416263662755, -0.008809228705477512, -0.1664858529644982, -0.12716971213834521, 0.006296288836362605, 1.2184949328315169, -0.7565783931654816, -0.8595323707527062, 0.24925744286410662, -0.23902891414240857, -0.05244624487938482, 0.2766614715127791, 0.1615303548990042, 2.0985367212757944, -1.7043719050719144, 0.056236528964586155, -0.35477077672605817, -0.016757479097702158, -0.012395512122671856], "ratio96": 1.9725848769471406, "ratio_final": 1.9726171510423114, "diag": {"ratio": 1.9726171510423114, "nonnormal": 1.3388737146635046, "eig_gaps": [0.12041467179846696, 0.17302527701617804, 0.17447643263561038], "discness": 0.007913642179095997, "peaks": 3, "deg": 3, "diamW": 1.7392785010015854}} +{"family": "nearnormal", "seed": 503010, "x": [0.22702796052230567, -1.8698423864234233, -0.29818029006658475, 2.2649786012211583, -0.35569683854275935, -1.4359238902281175, 3.545077379036972e-06, 1.536909083241135e-05, 0.1558874727010159, 0.1490695645278477, -1.476215756831183e-06, -1.6714741541188154e-05, -0.2841868412861729, 0.2319020915422329, 0.19878522901753773, 1.1602598807130757, 0.5697092793907033, 1.6460801018152058, 1.0710352577599354, 0.03824250540703592], "ratio96": 0.9999999999695982, "ratio_final": 0.9999999999783298, "diag": {"ratio": 0.9999999999783298, "nonnormal": 0.021012667300176053, "eig_gaps": [0.17338876608481182, 0.8833327990977008, 0.9947101204093693], "discness": 0.49701356770223437, "peaks": 1, "deg": 3, "diamW": 4.190209405764348}} +{"family": "companion", "seed": 504010, "x": [-0.8215443934040554, -0.6228458326789117, -0.33374409125174015, 0.3217756429767192, -0.4143272225473712, 0.9807316949213511, 1.0174919761974088, -0.21145016807395395, 0.44900535618942333, 0.0931431218542505, 0.903129876288792, -0.05311013154509401, -0.1124869797065793, 1.0512284611332907, -0.13131284932008014, 0.8738977090613216, 0.062071235256496965, 0.3277672309813259, 0.9293311315118112, -0.1939057579104183], "ratio96": 1.781569880435885, "ratio_final": 1.7811672748407623, "diag": {"ratio": 1.7811672748407623, "nonnormal": 0.6471454576253582, "eig_gaps": [0.30106267093954775, 0.48213216759978805, 0.7503040513085761], "discness": 0.3203597549444552, "peaks": 4, "deg": 3, "diamW": 2.205072398324683}} +{"family": "peaks", "seed": 505010, "x": [-0.8938077191292364, -0.03995844874084214, -0.09299238203621657, -0.24614483778035257, 0.8634686811648897, -0.7157701802341019, 1.863685041112002, -0.11689580949523709, -0.054130436146221844, -0.00998765436725401, 1.198968490747382, -0.7042189561422916, -1.7950439154054303, 0.1920939979283788, -1.0765951830817087, 2.3223057305486856, 2.607715939241275, 1.2370506358978002, 0.26659181747475125, 0.6184857474955132], "ratio96": 1.9391649252978824, "ratio_final": 1.9390347714661518, "diag": {"ratio": 1.9390347714661518, "nonnormal": 0.6973256228617007, "eig_gaps": [0.27721233700498793, 0.35719894722335194, 0.6311529056731684], "discness": 0.11413065497735517, "peaks": 4, "deg": 3, "diamW": 2.9830306796896435}} +{"family": "real", "seed": 506010, "x": [-0.30300505942341416, 0.05581254866580168, -0.4681059222307944, 0.0910934244737035, -0.31519384573231235, 0.07420613671943199, 0.10151333879643479, 0.224615332882677, -0.16392417831146894, -0.18639897430641034, 0.19978906018473996, -0.14112905485341176, -1.038891334716478, 0.3532666414999521, -1.4595864301067232, 0.13208684661047637, 1.4757366877989033, 0.4105837599662385, -1.5170092566579978, 0.8341108793759731], "ratio96": 1.9238737080722101, "ratio_final": 1.92459646692251, "diag": {"ratio": 1.92459646692251, "nonnormal": 0.43415553723606426, "eig_gaps": [0.04971525599146524, 0.3466156724318241, 0.38038163585860635], "discness": 3166252771.601639, "peaks": 3, "deg": 3, "diamW": 0.44383956319020795}} +{"family": "random", "seed": 500011, "x": [-1.473690819064787, -1.0990063480932855, -1.004065792241105, -0.7595595406262587, -1.6786200934563809, -0.9957916245033859, 0.6245556908987903, 0.6758564725816174, -0.6040257987983245, 0.08820986588481011, -0.32249841212297925, 0.4587070649573013, 2.09788538003055, -2.3858020390617813, 0.16941993573760072, 1.4098915687896403, 1.8854594944968994, 2.4497243830690696, 0.569310037007658, 0.17011387788199195], "ratio96": 1.946229689354994, "ratio_final": 1.9468717506582869, "diag": {"ratio": 1.9468717506582869, "nonnormal": 0.22725309765813118, "eig_gaps": [0.166230490628938, 0.4197943131528951, 0.5177883286335802], "discness": 27355515619.89297, "peaks": 4, "deg": 3, "diamW": 1.3803383653629968}} +{"family": "jordan3", "seed": 501011, "x": [0.2679801088664263, -0.6576042731060547, -0.4017001054304554, 0.4159664871473412, 0.05143550202829378, -0.09757391741934052, 0.7720539433018223, -0.568303051871085, 0.6977295984860811, -0.18546908660306427, 0.9691155709473218, 0.47505759888485, -0.02164931487862244, 0.5265925260153557, -0.18742609015410508, 0.32657545019513434, 1.6722828930719043, 0.4789659172663262, -0.6492078999707998, 0.271807529416459], "ratio96": 1.8676339146531065, "ratio_final": 1.8942031515403743, "diag": {"ratio": 1.8942031515403743, "nonnormal": 0.8486313441065582, "eig_gaps": [0.2927529933277705, 0.33392234781613395, 0.6169251443189715], "discness": 0.10819810592332903, "peaks": 4, "deg": 3, "diamW": 2.0510048595678487}} +{"family": "jordan2", "seed": 502011, "x": [-0.07291649359855701, 0.05069778413274992, -0.026673164111510023, -0.41272516733249964, 0.14242686631564067, -0.6632184216347453, 1.3312381233144934, -0.06973118257026525, -0.3212145785011361, 0.007034330944552764, 0.06460651799977152, -0.09537007915080364, 0.15281159844019687, 0.24566549209876798, 1.454700660179582, -0.5893191638429898, 0.10220907457652678, 0.015302873767812905, 0.07209428780719035, -0.042086791474647486], "ratio96": 1.9919452121747758, "ratio_final": 1.9919396766211583, "diag": {"ratio": 1.9919396766211583, "nonnormal": 1.109452243204337, "eig_gaps": [0.2086469093707921, 0.32151888387871025, 0.5147946306164923], "discness": 0.024791324642718007, "peaks": 4, "deg": 3, "diamW": 1.4485135664854916}} +{"family": "nearnormal", "seed": 503011, "x": [-0.9205846446704189, 0.998917286792752, 1.397950081278148, -2.5094662938852275, -0.6113718031148908, 0.5920468760546777, 1.2989118299948356e-05, -3.7256240554611837e-06, 0.09250258605115178, -0.07535377007866059, -7.037442821120625e-06, 1.9588421811929077e-05, -0.19071582508078275, 0.4151200316528447, 0.9138814384289267, 0.9286496460631006, 1.1973929689311482, 0.5816191581474008, 1.0506466144656166, 0.07789743267902545], "ratio96": 0.9999999999671868, "ratio_final": 0.9999999999789269, "diag": {"ratio": 0.9999999999789269, "nonnormal": 0.008171801821848188, "eig_gaps": [0.12132349232073236, 0.8773413898909357, 0.9983652638830948], "discness": 1.1436714048364587, "peaks": 1, "deg": 3, "diamW": 4.212163248354216}} +{"family": "companion", "seed": 504011, "x": [-0.5716366664945416, 0.6045527911611088, -0.41061333677342005, -1.8387077529433693, -1.2411734616367092, -1.6539199915564444, 1.1429305367984999, 0.2781900681762446, 0.3666930725918948, -0.29168350564045553, 0.9188761207951566, -0.04609510250764327, -0.07274282389224658, 0.03254605650524014, -0.316418392917439, -0.41177716411071685, 0.18581441598036957, 1.1603648858595639, 0.6421229654846909, 0.038335633365386146], "ratio96": 1.5387860370064235, "ratio_final": 1.5434588388660222, "diag": {"ratio": 1.5434588388660222, "nonnormal": 0.4324817373899727, "eig_gaps": [0.27923639213778334, 0.7730654712195839, 0.8035643414833571], "discness": 1.0738840594991081, "peaks": 4, "deg": 3, "diamW": 3.0471249129756526}} +{"family": "peaks", "seed": 505011, "x": [0.21813704121953295, -0.7819493841535656, 0.5154947384272697, 0.36343674656551894, 0.8768566979396772, -0.2362177304777452, 1.3958065019404873, 0.24160682135849756, -0.05785612244237755, 0.22035610220728366, 1.637425540137683, -0.5070337100503619, -0.9033672067614452, 1.4385094420822824, 3.8781445013585962, -2.1611347557309393, -3.562468405709703, 0.6829253551022583, 0.14124264771409764, -0.8029538651155212], "ratio96": 1.960840831472854, "ratio_final": 1.9608644336249832, "diag": {"ratio": 1.9608644336249832, "nonnormal": 0.6438351270590273, "eig_gaps": [0.2779337564457444, 0.3395827655729416, 0.46976882013238064], "discness": 0.05865335756338977, "peaks": 3, "deg": 3, "diamW": 2.5190173673163088}} diff --git a/problems/crouzeix/data/census_i6.jsonl b/problems/crouzeix/data/census_i6.jsonl new file mode 100644 index 0000000..480fb94 --- /dev/null +++ b/problems/crouzeix/data/census_i6.jsonl @@ -0,0 +1,79 @@ +{"family": "random", "seed": 600000, "x": [-0.483786559547726, 0.529573307564668, 0.5423554662736787, 0.700604448364845, 0.2003892865476582, -0.9379511441330397, 1.166475726748983, 2.558035428850487, -0.7440065067172372, 0.38525720604892466, 0.37862213557250685, -2.4026219241308846, -0.5123909882802855, -0.7327540619511552, -0.053774589053468186, 0.7516250019247344, -0.7494083019999997, -1.8625815128691303, 0.0386318040013974, -0.23974496576024928], "ratio96": 1.9818846153263343, "ratio_final": 1.981859301889913, "diag": {"ratio": 1.981859301889913, "nonnormal": 0.7459930417423575, "eig_gaps": [0.24735364397783494, 0.3849942548879538, 0.3979969004630632], "discness": 0.027865492762448325, "peaks": 4, "deg": 3, "diamW": 4.205709642683019}} +{"family": "jordan3", "seed": 601000, "x": [0.17013622931512906, 0.220448812761221, 0.005114668177192909, -0.33876147527108125, 0.33224063773806956, 0.2614122286241456, 1.249578181882702, -0.4932143040228865, -0.08949146000037389, -0.2623022628289994, 1.2549500045344388, -0.38484419248921137, 0.04871542618851671, -0.13278253016951969, -0.6369782854536392, 0.4215742543102404, 1.2511076363124454, -1.538119289897737, -0.06091589287167187, -0.18932046413297807], "ratio96": 1.9894734611921723, "ratio_final": 1.989449813565899, "diag": {"ratio": 1.989449813565899, "nonnormal": 0.6984908041111284, "eig_gaps": [0.08228814466136145, 0.2869507767408545, 0.33640438373228354], "discness": 0.02742435843436653, "peaks": 3, "deg": 3, "diamW": 2.031884455404855}} +{"family": "jordan2", "seed": 602000, "x": [0.40821028860113295, 0.026852762456371027, 0.6022418286059905, 0.12329976786161212, 0.5813308957274721, -0.22881317559643946, 0.9770874542161818, 0.12097579390298441, -0.10529858427872957, -0.1492703765558006, 0.08298078652475205, -0.0667404997573717, -1.1716190168587466, -0.37069708386950606, 1.971816620765022, 0.3335332259326218, 0.48543589800715004, 0.029920339133090806, -0.07386335418940909, 0.10318390672625373], "ratio96": 1.9818427901872062, "ratio_final": 1.9825064773241765, "diag": {"ratio": 1.9825064773241765, "nonnormal": 0.7475589862246607, "eig_gaps": [0.2104080463602655, 0.29982743730084205, 0.3425230546408032], "discness": 4751874114.8994, "peaks": 3, "deg": 3, "diamW": 1.029808974916182}} +{"family": "nearnormal", "seed": 603000, "x": [0.17560300293612, -0.8758908945058206, -0.8846560287947032, 0.2790137031652996, 0.6006684138364444, 0.13416260666577498, -1.3807821069054902, -0.6539790050433364, -0.283450992314791, -0.45539637026866453, 1.3227907227442048, 1.1125207926023055, -0.4033130964607201, 0.27507187561047325, -0.8483384009377757, 0.6257194812591425, 1.9458014465347626, 2.328222719650556, -0.3924442337197076, 0.756101231141817], "ratio96": 1.7855173202042904, "ratio_final": 1.7996378111063278, "diag": {"ratio": 1.7996378111063278, "nonnormal": 0.753456733493225, "eig_gaps": [0.38361276866868, 0.5224183410484214, 0.5488182450762725], "discness": 0.04080561542980108, "peaks": 3, "deg": 3, "diamW": 2.8566584690387633}} +{"family": "companion", "seed": 604000, "x": [-0.061057044337958555, 0.10831093193712528, 0.7161448110956626, -0.6433090132762234, 0.1294673342888505, -0.3804555954345675, 1.7385530025948759, -0.5536289050387474, -0.7379542789233176, 1.4368903197948648, -0.1268014088762755, -0.46160087865000216, 0.7756087761415511, -0.3538160998075639, -2.4134113395612022, -1.004226619284542, -0.09405303233132878, -0.2670455903110769, 0.046233547312059683, 0.08527236969806083], "ratio96": 1.8557367583138789, "ratio_final": 1.8609734670980096, "diag": {"ratio": 1.8609734670980096, "nonnormal": 1.1679040316395943, "eig_gaps": [0.1987212374073686, 0.24352840430749703, 0.4095706607167262], "discness": 0.02696145036959189, "peaks": 3, "deg": 3, "diamW": 2.639817006388698}} +{"family": "peaks", "seed": 605000, "x": [-1.3992271248326735, -0.4012883207271959, 0.49752713679492966, 0.29913697400388584, 0.8168948628427053, 0.7792848628386229, -0.0211200031532976, -0.020631587334709645, 0.012599224684066507, 0.0003119323865111042, 1.3627112661286647, 0.5176860258374449, 0.05773970963496267, 2.4983360815210283, -1.4228132366831063, -0.9022769802795925, -1.522020238865631, -1.259677459075421, 0.47320663672620067, -0.1749676393562627], "ratio96": 1.8361126677437953, "ratio_final": 1.8222197897543364, "diag": {"ratio": 1.8222197897543364, "nonnormal": 0.5519545844368837, "eig_gaps": [0.1902583286280776, 0.667103340601288, 0.8284456471017063], "discness": 0.35975411113759553, "peaks": 1, "deg": 3, "diamW": 3.030936187345354}} +{"family": "real", "seed": 606000, "x": [1.45320120177787, -0.013891352829364947, 2.0296394290376654, -0.002277076993149135, -0.4034640594559853, 0.01035145352354874, 0.9610140301140722, -0.5611309701090141, -0.697575870822283, -0.1963260956231899, 0.8226019228077479, 0.6690753570423664, 0.2045267341945219, 0.019920258877799853, -1.2861421866045128, 0.08350938947337666, 0.42479634958889745, -0.0001747953236459178, 0.11275950813625019, -0.03369594833746278], "ratio96": 1.8205397543588053, "ratio_final": 1.8214054701696325, "diag": {"ratio": 1.8214054701696325, "nonnormal": 0.4290727296076895, "eig_gaps": [0.19092983076300446, 0.6148986072508711, 0.8057481378677556], "discness": 0.7679564489364819, "peaks": 4, "deg": 3, "diamW": 3.0197230956385126}} +{"family": "random", "seed": 600001, "x": [-1.3033134754288342, 3.473470684357058, -1.1244929774053012, 3.3355258263837033, -1.3724416716476298, 3.492839620071009, -0.16924618424847737, -0.7048133450411078, -0.1604740362582054, -0.6136749558903021, 0.4292459602290378, -0.08729405457309519, 0.3539675105223247, 0.7601813939566148, 6.502550729283662, -1.4574504121284626, -1.0934595221238421, -4.530156447007874, 1.3613601012332595, -1.0508601932470936], "ratio96": 1.8549802017412118, "ratio_final": 1.854996819679032, "diag": {"ratio": 1.854996819679032, "nonnormal": 0.0372557468219564, "eig_gaps": [0.06677194676607481, 0.21005660006871799, 0.27311602356354747], "discness": 8320025814.095204, "peaks": 3, "deg": 3, "diamW": 1.0751582809203684}} +{"family": "jordan3", "seed": 601001, "x": [-0.10363451909761393, -0.27102920609892656, -0.1637355825105048, 0.6696457385389887, 0.7242292325480246, 0.21745793196418423, 1.6906500622286114, 0.6547665917668324, 0.018607580907323645, -0.031165020490362655, 1.5785889810096232, 0.8475985950010101, -0.02245939588893113, -0.0333091468646395, -0.17065767627852296, -0.2140539561843916, 0.5294367819331034, 0.016316177698056322, 0.045615336125179484, 0.08893786673681311], "ratio96": 1.9745921289102404, "ratio_final": 1.9746157296609834, "diag": {"ratio": 1.9746157296609834, "nonnormal": 0.7657367809859611, "eig_gaps": [0.34540419937906003, 0.3522364166937383, 0.3651474532270104], "discness": 0.012803501749542945, "peaks": 4, "deg": 3, "diamW": 2.7289562844117268}} +{"family": "jordan2", "seed": 602001, "x": [0.4703733981271889, 0.17602323463017222, -0.16319766013510872, 0.19558646468505225, -0.1740832501728684, -0.046147379095912704, 1.25971805957531, 0.6782201087559052, -0.1107482966304674, 0.013433261799948387, -0.021384370345143906, -0.012988988565902766, -0.201842161722655, -0.17973577167366867, 1.1586349583294293, -0.08563722617628447, 0.06881238065147963, 0.0903462963738606, -0.09114138422143683, -0.0020345386195672333], "ratio96": 1.984552682755901, "ratio_final": 1.9845199763927919, "diag": {"ratio": 1.9845199763927919, "nonnormal": 1.3212748496762168, "eig_gaps": [0.15424612227294726, 0.40405378076271703, 0.4345260711305998], "discness": 0.0463151630391169, "peaks": 4, "deg": 3, "diamW": 1.5687837868577623}} +{"family": "nearnormal", "seed": 603001, "x": [1.5171417610202846, -0.314308209807026, 2.1592002016991674, -0.8983639259723446, 0.4378371015782165, -1.1703854139437566, 1.4453998406491662, -0.9033338868668751, -0.13938242631533693, 0.06915962740152853, -0.27556558469150216, 0.03221817689522289, 5.607670186294116, 5.803184301685721, 3.5824484174565607, -7.77883454196755, -4.808512632376088, 1.3028887180205353, 0.6749335884664274, 0.20362779175707438], "ratio96": 1.7413819131845125, "ratio_final": 1.7640525690219728, "diag": {"ratio": 1.7640525690219728, "nonnormal": 0.3802811793069233, "eig_gaps": [0.36049528520543034, 0.5721628942153648, 0.7238138088419271], "discness": 122422490027.4814, "peaks": 3, "deg": 3, "diamW": 2.4076964057098773}} +{"family": "companion", "seed": 604001, "x": [-2.086126728161955, -1.8662579453717205, -0.06059696334083362, -2.5530890453583446, 2.0736391020752825, -0.644460916214421, 0.986738812711021, -1.1250644507622223, -0.6313582118774516, -0.2143536143034987, 1.10573696145924, -0.9608745786595287, 0.4594121758809362, 0.29620184571951286, 2.5833749182526535, -0.39563570663360914, 0.4942183466581358, 0.26233876160417813, 1.0186721146891733, 0.5644404239667464], "ratio96": 1.380750452255246, "ratio_final": 1.3853016218976717, "diag": {"ratio": 1.3853016218976717, "nonnormal": 0.36380431435456173, "eig_gaps": [0.4392573058339129, 0.5880254558585318, 0.8903988824862764], "discness": 1.1809878114992287, "peaks": 3, "deg": 3, "diamW": 4.869150430817396}} +{"family": "peaks", "seed": 605001, "x": [0.5075523671628827, -0.04077685050903449, 0.2739199379074265, -0.2547681366179897, 0.8869633908345655, -0.15584063867745312, 0.020000406074527438, -0.003521511219083296, -0.027049799649832798, 0.030892482765432196, 1.1489495227561095, -0.02070481291939345, 1.8714471069203182, -0.5194614532265609, -2.723992102676408, -0.5912250692226877, -0.9999774205534904, 0.67693197868297, 0.5297435914055975, -0.18295322875465747], "ratio96": 1.9699480337148254, "ratio_final": 1.9704061233923598, "diag": {"ratio": 1.9704061233923598, "nonnormal": 0.8391764379928809, "eig_gaps": [0.24240208713221226, 0.303344662379142, 0.47510994749862395], "discness": 27740797276.22095, "peaks": 4, "deg": 3, "diamW": 1.3070115190623237}} +{"family": "real", "seed": 606001, "x": [-0.3815346818587016, -0.162211618284429, -0.8334672516783328, 0.04268669202973023, 0.30425181912458943, -0.07419620983558725, -0.3495436053102085, -0.1831020152584926, 0.570722526267198, -2.032522804166162, -2.730300068476479, -0.7409386328794633, 0.7465685530222768, -0.24806636396525467, 4.23565055875034, 0.1833471801034355, 0.05715404116321318, -0.045862820702841596, -0.03090234514853768, -0.012043370145907787], "ratio96": 1.990399538287113, "ratio_final": 1.990434852368694, "diag": {"ratio": 1.990434852368694, "nonnormal": 1.3688252985267515, "eig_gaps": [0.13310747515979288, 0.18546918168868187, 0.3067962497457659], "discness": 0.021204701097621424, "peaks": 4, "deg": 3, "diamW": 3.7279049566377154}} +{"family": "random", "seed": 600002, "x": [1.6168180598408255, 1.0304230373511154, 1.1258948508526974, -1.1176309985805368, 0.8694173032349636, 1.6086483497604227, 0.11809189896994073, 1.7782359744623315, -0.9447811754758331, 0.7814795212288858, -1.7586295196440793, -0.5221822126698483, -2.5662433710206507, -2.880921400778191, 1.1299966145080003, 2.2855052766806985, -0.16188116958400536, 0.6604806883484793, -0.9213061301168428, -0.5488532631937661], "ratio96": 1.5326847834451718, "ratio_final": 1.5989092253951127, "diag": {"ratio": 1.5989092253951127, "nonnormal": 0.5815032511067285, "eig_gaps": [0.2353978807004419, 0.5488950950282966, 0.682137791760927], "discness": 0.3209466746461186, "peaks": 3, "deg": 3, "diamW": 4.014316463835713}} +{"family": "jordan3", "seed": 601002, "x": [0.4910859889433299, -0.5406588116263269, -0.14370609615466085, 0.004371559501834217, -0.5634969810799144, 0.23181768057059615, 1.1650093050227408, -0.5722097351445888, 0.21033693109984034, 0.04829068608793158, 1.2615093689536807, 0.020630260335673868, -0.0024845885977623624, 0.3409444791336388, 0.33275778132387335, 0.23588267067740248, 1.414047809357368, -0.40207553019282566, -0.1459259817764797, 0.02366637231059926], "ratio96": 1.9703536012156118, "ratio_final": 1.970341072659899, "diag": {"ratio": 1.970341072659899, "nonnormal": 0.6951641486102499, "eig_gaps": [0.21237452306232957, 0.3721617434121056, 0.581474511962557], "discness": 0.08370423617935056, "peaks": 4, "deg": 3, "diamW": 2.2481392936291766}} +{"family": "jordan2", "seed": 602002, "x": [-0.08855164460637693, -0.19335783108193771, 0.44139881227160294, 0.2087842991615908, 0.4580486509190249, -0.18558394670997136, 1.543179191826876, 0.4915964812787917, 1.3270035046667061, 0.4101281114578845, 0.05353399466636341, 0.1624163897077251, -0.2795943186190175, 0.1423045742338035, 1.329448510615379, -0.31451427212993044, 0.032566710083742303, 0.002083974443680742, -0.0017838683005976354, 0.02075566429182786], "ratio96": 1.9930090616710385, "ratio_final": 1.9930091694590535, "diag": {"ratio": 1.9930091694590535, "nonnormal": 1.3333518076396678, "eig_gaps": [0.17628378837349462, 0.24413919507138884, 0.29710683491840184], "discness": 0.01505245952878435, "peaks": 3, "deg": 3, "diamW": 2.2391143446750967}} +{"family": "nearnormal", "seed": 603002, "x": [0.6601748067002443, -0.3852464793174133, 0.5418962584423367, -0.24186584921444648, 0.691228780783226, -0.36819393026470926, 0.23708897171391075, -0.1514582651326482, 0.4597487272978425, 0.005316828414408967, -0.11581094322081892, 0.10750526541881955, -0.3123960418724254, -0.7938539187159114, 0.6070299079980277, 0.4246699897177282, -1.8452587320587757, 0.4969921335825642, 0.6989074691079303, 0.2861574784839396], "ratio96": 1.9693415847322253, "ratio_final": 1.9718211928815372, "diag": {"ratio": 1.9718211928815372, "nonnormal": 0.252494782926166, "eig_gaps": [0.058742958064697266, 0.3081912258130484, 0.32432219832518666], "discness": 7608349979.139288, "peaks": 4, "deg": 3, "diamW": 0.6031010079411856}} +{"family": "companion", "seed": 604002, "x": [-0.5720987789816994, 0.21343716876010665, 2.174800486950332, 0.015844027870660568, 0.08288124358921045, -0.1666925739451262, 1.4745224360668778, -1.6762965623304682, 0.9562328585367572, -1.6511339225553354, -2.014858773720798, 0.9298392462524004, -0.6605585458139289, 2.4221014130959535, 0.024518344538230777, 5.809574643793576, -0.3355020491410424, -3.868984646818368, 0.30800353932527325, -0.3440349168978257], "ratio96": 1.960300814195675, "ratio_final": 1.96025972749273, "diag": {"ratio": 1.96025972749273, "nonnormal": 0.6875588886467539, "eig_gaps": [0.1625469072829485, 0.4507180655461723, 0.5911210203469861], "discness": 0.10650156123378811, "peaks": 4, "deg": 3, "diamW": 4.658939119018658}} +{"family": "peaks", "seed": 605002, "x": [-1.4925882664586005, 0.22958050930881915, -0.30763699049448157, 0.24029363769638973, 1.143469430973672, 0.09776819068140394, 1.8646257084767965, -1.1146877796814072, 0.43959294982178226, 0.1255937373786603, 2.572862822588974, 0.5829737647250971, 1.6906166911383063, 0.11922130696511529, -2.068431648577338, 1.1635896619729886, -2.635451599414301, 0.232987259686058, 0.38348271450930105, -0.0027854281294325306], "ratio96": 1.9684686596695675, "ratio_final": 1.9702663383359411, "diag": {"ratio": 1.9702663383359411, "nonnormal": 0.7238337601804495, "eig_gaps": [0.27452432021402057, 0.33778985188400573, 0.6114483297708638], "discness": 0.0860870436540157, "peaks": 4, "deg": 3, "diamW": 4.316556371671547}} +{"family": "real", "seed": 606002, "x": [0.1195065285189599, 0.21281077765666173, 0.27671792554689356, -0.024608702079277435, -0.3185470749662682, -0.05573117268874665, 0.3645802000314601, 0.08233915712455124, 2.2046211739807435, 0.5879001027687792, -0.5055032989779731, 0.17670315091212666, -0.4510536339405876, 0.2236613057483524, -2.2946198968908362, 0.49490133216288523, 0.10487045411951756, -0.09460508442749552, -0.005665638640628461, -0.012697009269545305], "ratio96": 1.9858874406934037, "ratio_final": 1.9879277920417806, "diag": {"ratio": 1.9879277920417806, "nonnormal": 1.370120407306207, "eig_gaps": [0.11759881303826007, 0.21219911025831287, 0.24617276119014245], "discness": 0.010812234155000792, "peaks": 3, "deg": 3, "diamW": 2.421380970351202}} +{"family": "random", "seed": 600003, "x": [0.04790071821568607, -1.1346993324158836, 0.33685352483381237, -0.7878878945287856, 0.1069758807235365, -0.09010455946111502, 0.2582944118462696, -0.7152691015907005, -0.07417576466374047, 0.4301075111850337, 0.8764357779664764, -0.5177489961251561, -1.1343909380591433, -0.3830731987784674, -0.13370011236780474, 3.019055114930917, 0.4576524345897668, -0.33575467465521547, 0.0688903303364937, 1.3477796628455714], "ratio96": 1.872437307991583, "ratio_final": 1.8970449526308026, "diag": {"ratio": 1.8970449526308026, "nonnormal": 0.5110848592249776, "eig_gaps": [0.2660734015210353, 0.43303575500863006, 0.6166951750809605], "discness": 0.36871219312864045, "peaks": 4, "deg": 3, "diamW": 1.6965657038684547}} +{"family": "jordan3", "seed": 601003, "x": [0.14409124557984498, 0.01290867607529127, -0.2476299908580556, -0.15265270808686954, -0.2656237136393142, 0.2971042870153203, 1.682474274192485, -0.001903370764457356, 0.31839147840455834, 0.09862497431122462, 1.3526341931595554, 0.9268646512143232, 0.035146139043216826, -0.06271585597806621, 0.29667123709006327, -0.39146340987576755, 1.2286730176438527, -0.8049429076784971, -0.1749605658234433, 0.04867772760466256], "ratio96": 1.9948218841283025, "ratio_final": 1.994816612552754, "diag": {"ratio": 1.994816612552754, "nonnormal": 0.7182665573214544, "eig_gaps": [0.17581791556602952, 0.1860894553524695, 0.2061466886692669], "discness": 0.0020367420389539167, "peaks": 3, "deg": 3, "diamW": 2.4188194525083984}} +{"family": "jordan2", "seed": 602003, "x": [0.5153699363202136, 0.19462389284054138, 0.3573371927568145, -0.15232418478818666, 0.3400632757232717, -0.2952172375032175, 0.8224801157559094, -0.32885844908902384, 0.24769124547840723, 0.41133132948321616, -0.16850143448364185, 0.021352286743649404, -0.2221471973357515, -0.4887496287208082, 0.46751758606462124, 0.819849261297164, 0.32112337245978306, 0.07681885291822706, 0.0028888809732940854, -0.012660217472187148], "ratio96": 1.9262834262497317, "ratio_final": 1.9316633108665184, "diag": {"ratio": 1.9316633108665184, "nonnormal": 0.8993388479980872, "eig_gaps": [0.133636299306548, 0.35397001935959044, 0.4830459594618915], "discness": 0.11412274924062075, "peaks": 3, "deg": 3, "diamW": 1.0770528878878565}} +{"family": "nearnormal", "seed": 603003, "x": [-0.6924544199634555, 0.19307787059626738, -0.3228443835266245, -0.963790739303729, -0.21966786000434335, -1.0069693356746352, -0.028835706002084634, 0.01461277912570846, 0.004704060735565471, -0.04061763174301851, -0.2737481642795961, 0.173050301058706, -0.11230452190153656, 0.1831602076179899, 0.12890193790651955, 1.6050918792666096, 1.323595276341583, 1.7293634717967015, 1.2591644683868632, 0.34084754650970384], "ratio96": 1.719009451637423, "ratio_final": 1.7187594873165428, "diag": {"ratio": 1.7187594873165428, "nonnormal": 0.06719854514291566, "eig_gaps": [0.07884619484016314, 0.8561412258618379, 0.9092546446864818], "discness": 187528502750.3034, "peaks": 2, "deg": 3, "diamW": 1.418548551575718}} +{"family": "companion", "seed": 604003, "x": [0.6323854176586391, 1.3007272389119027, -0.8901792033258537, -1.211673903641992, 1.3410040970172952, 0.06532475771371864, 6.313453073383169, -3.5666102411055536, -1.5666785905773482, -0.7487794628090111, 0.32391473487867206, 0.008852522752009907, -0.23664681590875725, 3.833933866035115, -6.048535569459153, 0.15038548532386964, -0.011532710474450314, 0.19569355005625533, 0.002141111404760797, -0.009348803527375237], "ratio96": 1.943821970391539, "ratio_final": 1.9462486629760212, "diag": {"ratio": 1.9462486629760212, "nonnormal": 1.3603102259622077, "eig_gaps": [0.1770560549096928, 0.31959724890644126, 0.36521853798192055], "discness": 0.027447398942766495, "peaks": 3, "deg": 3, "diamW": 8.043808188303133}} +{"family": "peaks", "seed": 605003, "x": [-0.8927896417298795, -0.10143110142024458, -0.6643006988119424, 0.19513225065728135, -0.047450026984793475, -0.262837344669258, 1.1368341603378456, -0.12098354743315333, 0.4001210386490157, 0.1591658163989167, 1.2645873535874668, -0.18803253331106912, 0.8219215102112075, 1.2969157341758375, 4.066955999288061, 3.0148679887668344, 2.8880705650595426, 1.625775116436719, -1.203904006457277, -0.4385201991653698], "ratio96": 1.9482314751347063, "ratio_final": 1.9504934396249074, "diag": {"ratio": 1.9504934396249074, "nonnormal": 0.60640588469911, "eig_gaps": [0.18952161791740535, 0.3889251072734205, 0.43567049380474654], "discness": 0.06345218243591699, "peaks": 3, "deg": 3, "diamW": 1.9753709852752992}} +{"family": "real", "seed": 606003, "x": [-0.2566913770005327, 0.03475889703430658, -0.4265388424122325, -0.029481609145788693, -0.2559070395181571, -0.01077550647922096, 0.1197575394817135, -0.12516671381894157, -0.1980977003290842, 0.15300252886338317, 0.29661007428315905, 0.19236383008160923, -1.609952808190609, 0.300473477120563, -2.6355086688001985, 0.7205942225181977, 2.5684039858436325, -0.7440482499639566, -1.7943581229892072, 0.39813620364895624], "ratio96": 1.9284490194505288, "ratio_final": 1.9289403584145925, "diag": {"ratio": 1.9289403584145925, "nonnormal": 0.601821963989707, "eig_gaps": [0.09286620456983961, 0.3500320480904181, 0.3702934708886298], "discness": 4365565073.440474, "peaks": 3, "deg": 3, "diamW": 0.49039538542162986}} +{"family": "random", "seed": 600004, "x": [0.2731449129121549, -0.3500702624419034, -0.6573614232027121, 0.8320932219699912, -0.16246777743518864, 1.3657054533125472, -0.44282512979478544, -1.7800595586247796, -0.7200570649885343, 0.3713423840075123, -1.039382051891872, -0.8644479389560134, 0.108941250669809, -0.3425742722261296, 1.5211945338648252, -0.27699368388468393, 0.2898597818059422, 0.7415428800785169, 0.1424852130194656, -0.2613437681483193], "ratio96": 1.9347212377287883, "ratio_final": 1.9349931508332516, "diag": {"ratio": 1.9349931508332516, "nonnormal": 0.7023757408064905, "eig_gaps": [0.24401368949716779, 0.5044184874420107, 0.5935260307176465], "discness": 0.09189893234885027, "peaks": 4, "deg": 3, "diamW": 2.9825319691898393}} +{"family": "jordan3", "seed": 601004, "x": [0.30097624973635306, -0.05041173849327954, 0.10018271146439672, 0.04495547149506737, 0.08506834075734643, 0.2170800001336809, 1.6086043338609135, 0.2591359940466885, -0.12448829423257406, 0.12820382686810083, 1.6123930833458768, -0.014167380683331223, 0.003901834878486419, 0.09672017407290046, -0.38205719883502653, -0.40954762626773467, 1.3200504852509873, 0.6236482879770664, 0.15273029596051396, -0.04462918478146716], "ratio96": 1.9952131413435652, "ratio_final": 1.995203391499658, "diag": {"ratio": 1.995203391499658, "nonnormal": 0.716538156199698, "eig_gaps": [0.0743162227720351, 0.09560779804640068, 0.14785055878771358], "discness": 0.005559526964204263, "peaks": 3, "deg": 3, "diamW": 2.325022039919355}} +{"family": "jordan2", "seed": 602004, "x": [0.2053750906049205, 0.0984773580689698, 0.3270816828526707, 0.1713600972420975, 0.3008895534780897, 0.27808200848351106, 1.184327514954918, 0.011832699294115635, -0.04092432970748106, 0.2660205160723768, 0.023738699977539492, 0.016751884942643288, -0.37629028595993685, -0.24458395328426563, 1.355185096915448, 0.19196923022714854, 0.1017457866643933, 0.014654339668854847, -0.0020744639488088287, 0.01324687498673336], "ratio96": 1.996692475989796, "ratio_final": 1.9966914813962127, "diag": {"ratio": 1.9966914813962127, "nonnormal": 1.1470378927922793, "eig_gaps": [0.08982265697436022, 0.11595593714703796, 0.16627666816166164], "discness": 0.004930178870139324, "peaks": 3, "deg": 3, "diamW": 1.2233996549532116}} +{"family": "nearnormal", "seed": 603004, "x": [-1.233804600737956, 1.434234790212786, 1.7798159927375434, -0.7711993391473615, 0.4893245145874551, 2.195832340277118, -5.514007770573618e-06, 1.501263011935098e-05, -0.10906481812564148, -0.048332070135748334, -5.506079061766619e-06, 4.5498893965717527e-07, 0.3988553690165122, 0.7742252331025311, -3.044800532973633, 0.9142560411014835, -0.2521420665474901, -2.976458866236585, 0.8868684329476836, 0.11682891406881923], "ratio96": 0.9999999999708791, "ratio_final": 0.9999999999855295, "diag": {"ratio": 0.9999999999855295, "nonnormal": 0.025647633594428662, "eig_gaps": [0.5039699034569374, 0.8655343623809016, 0.9989909382798533], "discness": 0.7027323326576764, "peaks": 1, "deg": 3, "diamW": 3.7381861753118653}} +{"family": "companion", "seed": 604004, "x": [-0.9783212171591003, -1.0891028062411447, -1.3415849241803706, 1.2836160748262473, 1.7669446918734495, -1.0866782798562489, 1.0295226493508403, 1.7801751493193585, 0.644908815512653, 1.497295727925965, 1.5589510990177662, -0.6777479045453649, 0.23514093605357234, -0.22775789633578267, -0.18238405912874045, 0.06685600898848872, -3.4476573027790556, -0.29796630678873653, 0.5008223468861122, -0.21563136018131468], "ratio96": 1.5050741334613045, "ratio_final": 1.5758094914744551, "diag": {"ratio": 1.5758094914744551, "nonnormal": 0.6669731133671148, "eig_gaps": [0.47992008331853997, 0.5488783510163235, 0.781575576799967], "discness": 0.13101225153041857, "peaks": 3, "deg": 3, "diamW": 5.00159456932328}} +{"family": "peaks", "seed": 605004, "x": [-1.1278022501950402, -0.2545004242985418, -0.49831145644052044, 0.3959312904274337, -0.21415244657087218, -0.7684961500979393, 2.2490660018669537, 0.8247630737866214, -0.2873812977259188, -0.05858476607105431, 1.7122072006442177, 1.5531431266584623, 1.5293132299836891, 1.5931017502737865, 5.728348122426677, 1.7413248989161576, 4.339422109183824, -0.10410672036863039, 0.12597444124508556, 0.42926008312830444], "ratio96": 1.9785357609598182, "ratio_final": 1.978559453882369, "diag": {"ratio": 1.978559453882369, "nonnormal": 0.659681817852658, "eig_gaps": [0.2552931595531666, 0.29566564656698296, 0.33805393508673287], "discness": 0.01595833756472086, "peaks": 3, "deg": 3, "diamW": 3.545582700820795}} +{"family": "real", "seed": 606004, "x": [2.008530672267221, 0.7001503153799697, -1.2625604631327683, -0.7273530040644184, -0.4718457718448178, -1.1889037151020156, 0.38049367060081307, 0.6650160971787192, -0.7953869953075838, 0.4483143658894968, -0.013659823893765834, 1.0366196139922388, 1.2699382788615474, -1.0835473068251718, -0.29482541003532187, 0.1949345750478072, 0.7899334790728734, 0.8522037278132089, -0.14847195757798076, -0.23731752292111163], "ratio96": 1.5330881839246997, "ratio_final": 1.5463886315837072, "diag": {"ratio": 1.5463886315837072, "nonnormal": 0.5067020335211357, "eig_gaps": [0.22788154623029994, 0.7760164661060669, 0.8883157547038528], "discness": 0.37033360603735577, "peaks": 3, "deg": 3, "diamW": 4.01772214117389}} +{"family": "random", "seed": 600005, "x": [0.48028691300790793, 0.29739542156791327, -0.802802739046464, 0.4327438149368178, 0.8262234040599581, -0.9219955483887792, 1.6402557552396264, -0.005957361847453296, 0.5817625276278966, -1.1999467837332773, -0.43176083211167154, -1.8518951111512871, 1.5709262823747234, 0.6918519743210033, -0.22568516899566776, 0.6113671666541207, 0.2225010720173538, -5.493720402867936, -0.7142547389157953, 0.5107630574576634], "ratio96": 1.9224561116401375, "ratio_final": 1.93390661994052, "diag": {"ratio": 1.93390661994052, "nonnormal": 0.7560016273359537, "eig_gaps": [0.36011756482072305, 0.3665660133607969, 0.6019622238356217], "discness": 0.07718849185434486, "peaks": 4, "deg": 3, "diamW": 3.519716933229399}} +{"family": "jordan3", "seed": 601005, "x": [-0.21581488856420505, -0.2312006366628763, 0.3342442458054502, 0.12860888042082425, -0.15295109200542412, 0.43598757321482934, 1.0613697873143382, -0.6585645854574628, -0.48259362952924056, 0.23313805488615583, 0.9808945584988993, 0.537445429814906, -0.026119201572154582, 0.04689712031222895, -0.11154857086891568, -0.28327840866440523, 0.8182252098374758, -0.36801308920690623, 0.04634994813066974, -0.2757236319412787], "ratio96": 1.9797224204475175, "ratio_final": 1.9799329149467628, "diag": {"ratio": 1.9799329149467628, "nonnormal": 0.7473096235186139, "eig_gaps": [0.30740362785594216, 0.35075180242724496, 0.3576116318727301], "discness": 0.007350929189456908, "peaks": 4, "deg": 3, "diamW": 1.8739413987514382}} +{"family": "jordan2", "seed": 602005, "x": [0.09003171710320756, -0.11999367405616038, -0.04972393654309234, 0.0649696553336624, -0.11304217890155073, -0.3102651707193421, 1.1701023271671742, -0.4319496726978295, -0.38033155215520764, 0.28180840616016983, -0.17372685736547455, -0.14435820549201978, -0.03396057984802256, -0.02368905940648666, 2.0018389465839466, -0.44668814322050787, -0.1255790252471154, -0.12956955082588978, -0.003755564314191627, -0.04302407488400947], "ratio96": 1.9848054989623232, "ratio_final": 1.9848139471528905, "diag": {"ratio": 1.9848139471528905, "nonnormal": 1.3115284842021386, "eig_gaps": [0.16864122859460587, 0.2024377390287048, 0.2768231053461325], "discness": 0.010463682004651162, "peaks": 3, "deg": 3, "diamW": 1.37466698409107}} +{"family": "nearnormal", "seed": 603005, "x": [-0.4615300626789711, 1.0890592402493162, -0.18722824674137362, 0.10046490345884984, 1.0110325749667473, -0.498047744856593, 1.3319350599779447, 1.5527657059519153, 0.21324687129207498, 0.23547370873430024, -0.41246584466850045, -1.839371064046242, -0.8654448594187776, -0.43456796940202114, 0.6144858499739739, -0.051039055165771396, -0.5516101397678659, 0.9896464771867675, 0.013068680067292476, 0.18468721726382697], "ratio96": 1.9572253266066792, "ratio_final": 1.961538994240155, "diag": {"ratio": 1.961538994240155, "nonnormal": 0.7274247725660782, "eig_gaps": [0.2893917934557761, 0.3778153191033986, 0.610697846255246], "discness": 0.09148055244828178, "peaks": 4, "deg": 3, "diamW": 3.545171676464817}} +{"family": "companion", "seed": 604005, "x": [-0.335436846829379, -1.7394397732873117, 0.7287195336855687, 1.7602315418217267, 2.9027513426756775, -1.131686202035291, 2.403590405312177, 1.0669907106789012, 0.8514686762105939, 0.09393232459041548, 0.23345850708455784, 0.17456354018869502, 0.9414431809590948, -0.8008476742602169, -4.428177584449417, 1.7434297409729322, -0.8419037108396537, -0.16476056474587775, 0.1991341105072167, 0.13307823508723526], "ratio96": 1.382959740674868, "ratio_final": 1.4635977531638367, "diag": {"ratio": 1.4635977531638367, "nonnormal": 0.7518629804715131, "eig_gaps": [0.7161557599392148, 0.7864140449756183, 0.7950934344633599], "discness": 0.13468952070588447, "peaks": 3, "deg": 3, "diamW": 4.600573448887192}} +{"family": "peaks", "seed": 605005, "x": [-1.627479847110239, -0.44907584793851313, 0.36667760032044927, 0.0669178157957378, 1.1842336505033844, 0.30832263671788385, 0.6544949809742306, -0.24503301564341223, -0.12613521381345186, -0.025074256327523428, 1.8185328179291744, 0.08901639042234884, 1.5024878173197498, 2.189734413684656, -1.4398327683733747, -1.2118101475453684, -2.2159117714895036, -1.1100826253572693, 0.6169095746915572, 0.14696940247453338], "ratio96": 1.8269455289368668, "ratio_final": 1.8420652209399841, "diag": {"ratio": 1.8420652209399841, "nonnormal": 0.6384423064458211, "eig_gaps": [0.23874665202665238, 0.5768986302941969, 0.8155481530906135], "discness": 0.23314710751088577, "peaks": 5, "deg": 3, "diamW": 3.5705290864749113}} +{"family": "real", "seed": 606005, "x": [0.23987281551422526, 0.3732522286416609, -0.11903510842712742, -0.23539821630134355, -0.19155447714502857, 1.2418454134945716, 0.04416553946799092, 0.16034823533391393, -0.9145589154594972, 0.1038836688359851, 2.9784996037081632, -0.31760326843155634, 0.18397015118884677, -2.709166571238137, 3.6511369665963773, -1.793958814283818, -0.31340318406686973, -0.06417194565381498, 0.07707734242861607, -0.046898270878417934], "ratio96": 1.9396567513890708, "ratio_final": 1.944038870107737, "diag": {"ratio": 1.944038870107737, "nonnormal": 1.3170615683170301, "eig_gaps": [0.20369750994558938, 0.27958674534164135, 0.4263759261860648], "discness": 0.051035558074515104, "peaks": 3, "deg": 3, "diamW": 3.468822914301325}} +{"family": "random", "seed": 600006, "x": [-0.05095443983090295, -0.3607699703275107, -1.0622174445718864, 0.14920320973202159, -1.2184289258546386, 0.1904118445911489, 1.251366980958539, 0.08173938024498904, -0.41080683233957666, -0.7748136479309472, -0.10715713866567282, -0.0067463024039094086, 0.8635907270107408, 0.5829281140897684, 1.8254525307192297, 0.5265548675340221, 0.030096894333709018, -0.4106358024525291, -0.08529102670053346, -0.26148312494804715], "ratio96": 1.911470875578101, "ratio_final": 1.9153715426347542, "diag": {"ratio": 1.9153715426347542, "nonnormal": 0.8030130736600809, "eig_gaps": [0.08404877630480158, 0.5892187006007545, 0.6716625756644881], "discness": 0.2438831112923686, "peaks": 4, "deg": 3, "diamW": 1.9221635777617057}} +{"family": "jordan3", "seed": 601006, "x": [-0.058272399290189625, -0.24667086570729477, -0.5056371643022823, 0.10861159309549395, 0.5530049647880688, 0.09669716406117357, 1.8910284432628248, -0.4196952819142642, 0.5320155302718628, 0.20597203563423938, 1.7786414912184654, -0.3041298788375572, -0.19762810140114728, 0.06875203943871755, -0.01576694914713178, -0.04540178365528019, 1.193086481751724, -0.34010404988762577, -0.015958902613418535, -0.14112411013900394], "ratio96": 1.9967622641143539, "ratio_final": 1.9967694422831415, "diag": {"ratio": 1.9967694422831415, "nonnormal": 0.7262079146851976, "eig_gaps": [0.1963953942809951, 0.24103008207754498, 0.36396442879098645], "discness": 0.02453738116134489, "peaks": 4, "deg": 3, "diamW": 2.9088259411067434}} +{"family": "jordan2", "seed": 602006, "x": [-0.1506024507409685, 0.40914758132244566, -0.002420090669959518, 0.32131974876931735, -0.1687762084565399, -0.03938403582392612, -0.016401330916424938, -0.18726583007842618, -0.885431664693575, -0.6916434265268065, 0.3318271640510078, -1.251433461149873, 0.169534319470325, -0.10243654801744398, 1.4618716408187167, 0.6148421760491166, -0.013422911406622887, 0.08139088718810736, 0.002595054926757693, 0.028550100840205227], "ratio96": 1.9930868777421036, "ratio_final": 1.993105993398295, "diag": {"ratio": 1.993105993398295, "nonnormal": 1.3056979898087209, "eig_gaps": [0.09724428195753732, 0.2242439393712989, 0.2534205531844505], "discness": 0.012580972357830272, "peaks": 3, "deg": 3, "diamW": 1.7713624478525372}} +{"family": "nearnormal", "seed": 603006, "x": [0.6806268242780293, 0.09231473659106554, 0.3057122064111115, 0.1204532306247112, 0.5499176738545962, -2.883504777139101, -0.6301183386572278, -0.2312969224665807, 0.10504234360177234, -0.06786019736996708, 0.07606461459652447, -0.029647648054681206, 1.6359228526233929, 1.743349170045624, -2.87755640610899, -3.9379571676150906, -1.6909557689230361, 3.276671009011423, 0.7516185441950949, 0.18335164369143725], "ratio96": 1.7037919884708002, "ratio_final": 1.605585731404586, "diag": {"ratio": 1.605585731404586, "nonnormal": 0.09950905800750735, "eig_gaps": [0.11293950794617383, 0.8947854058556918, 0.9053530882057428], "discness": 440027496798.232, "peaks": 2, "deg": 3, "diamW": 3.328942033456778}} +{"family": "companion", "seed": 604006, "x": [0.5225441605490881, -1.7613118650685273, 0.5889496596213541, -1.4145887059457458, -0.4555675517706028, 0.9163087468313333, 0.8630803906648437, 0.3616335483373272, 0.0287224204908321, 0.011138946787502903, -0.02507645986131749, -0.00996487455186671, 1.696532736056071, -0.7922295732070382, 0.93991867984463, 0.6074034330103233, 0.1752653487754826, -0.4432628126777405, 0.3915871740211978, -0.00213731260614041], "ratio96": 1.7608217278803144, "ratio_final": 1.6609233832027015, "diag": {"ratio": 1.6609233832027015, "nonnormal": 0.17436249184411418, "eig_gaps": [0.11059798413867107, 0.800206445229223, 0.8930785433713792], "discness": 0.8532135470866169, "peaks": 1, "deg": 3, "diamW": 3.191965810420953}} +{"family": "peaks", "seed": 605006, "x": [-1.1130529289735753, -0.42256868589262486, 0.06865363037388733, -0.28214166672309005, -0.42574303272815783, 1.2680048604354237, 1.130126793657418, -2.1441003054207544, -0.14653098866284386, -0.3174489608639708, 2.7190705863778666, 0.3109041739793869, 2.192517909994469, -1.5243217406338299, 3.4135845831105303, -1.6543503295729602, 3.0137166931889006, 0.37097690285095586, -0.3396156462311285, -0.19936288286293907], "ratio96": 1.9529974794233007, "ratio_final": 1.9532009263603356, "diag": {"ratio": 1.9532009263603356, "nonnormal": 0.7388439986634598, "eig_gaps": [0.2884495705104577, 0.3943878967982939, 0.44234961146453], "discness": 0.03801425528039115, "peaks": 3, "deg": 3, "diamW": 4.125577531573073}} +{"family": "real", "seed": 606006, "x": [2.194078930974954, -0.16133264171300518, 1.8663653863386307, 0.04018776700677672, -1.3028342598008893, 0.07698918129032617, 0.8803405597072707, -0.09595124729347422, -0.10733887224304021, 0.20660247493305264, 0.22205517293805097, -0.04835076332298455, 1.8178047118634726, 0.10085545124698087, 1.5708551114283686, 0.19023216332283968, 1.0030627804084804, -0.01556701391891098, -1.0045349880988765, -0.1259896812145136], "ratio96": 1.6876085032318193, "ratio_final": 1.6270231550005925, "diag": {"ratio": 1.6270231550005925, "nonnormal": 0.17229299544190277, "eig_gaps": [0.10071952536305223, 0.8297589486133367, 0.9176227405690216], "discness": 0.7429776497061936, "peaks": 1, "deg": 3, "diamW": 3.819679579380595}} +{"family": "random", "seed": 600007, "x": [0.19050876180930423, -0.33931874149884944, -1.450342740785822, 0.11763006911247292, -0.147760397844507, 0.37224454316947386, 0.09076127105878186, -1.7553425230938853, 0.2733228254745358, -1.6220548717636376, 1.7762550993899886, -0.24035271422192084, -0.0172844485381676, 0.46773749840058526, 1.4560867387536074, 4.003179114732218, 0.9959617126324976, 2.987779312369618, -0.1694745131050691, -0.8516145209982313], "ratio96": 1.884040897673235, "ratio_final": 1.8869796101928276, "diag": {"ratio": 1.8869796101928276, "nonnormal": 0.7581064592742784, "eig_gaps": [0.22992967111891272, 0.3873331227844364, 0.4970793966779462], "discness": 0.04892745309412353, "peaks": 3, "deg": 3, "diamW": 3.4265948881404276}} +{"family": "jordan3", "seed": 601007, "x": [0.6602763872159385, -0.04212155390755326, -0.38148330407046827, -0.2495819158142698, -0.29484198858187277, -0.26671482927774925, 1.6008855898268655, 0.3317634854552758, -0.20671584687138825, 0.6338557025317748, 1.8741680282116966, 0.28830410037831444, -0.4187311646252495, 0.017499290805793745, 0.11757974688769009, 0.553090682330313, 1.2532147189119085, 0.08931471803867097, 0.04464402086294145, -0.19521368363255243], "ratio96": 1.7493161961282113, "ratio_final": 1.8974921440997967, "diag": {"ratio": 1.8974921440997967, "nonnormal": 0.8476050672494106, "eig_gaps": [0.03119361581858506, 0.3465415412889638, 0.37516664649464143], "discness": 0.05769357681048524, "peaks": 3, "deg": 3, "diamW": 2.831318071765534}} +{"family": "jordan2", "seed": 602007, "x": [-0.1755971928768279, 0.030048183450366556, -0.14410604817174472, -0.0035746936720660186, -0.19954201725929563, -0.08346223405702496, 1.2678346521186104, 0.3174409369367081, 0.027628912172930672, 0.1524371130868865, -0.004474898098319281, 0.015771080200343637, 0.11848530204303681, -0.1906390403508627, 0.6907080965425434, -1.126135678052263, 0.02336367605060407, 0.04943586533287095, -0.003238865062849223, -0.0008477517386126752], "ratio96": 1.9987495711606287, "ratio_final": 1.998750459681895, "diag": {"ratio": 1.998750459681895, "nonnormal": 1.3382640387073814, "eig_gaps": [0.034977466959736025, 0.07382964680278367, 0.08808176057979718], "discness": 0.0003257586144565633, "peaks": 3, "deg": 3, "diamW": 1.317054610838134}} +{"family": "nearnormal", "seed": 603007, "x": [-1.0513391630496558, -1.6959262119045477, 1.7186235004831798, 1.7915237366076406, -0.8367520905874531, -0.38497179063937015, -1.1961018030259693e-05, 8.96186592711346e-06, -0.021760220352189514, -0.1533253353863978, 1.0933630902174662e-05, -1.5484994447905858e-05, -0.013340490229108075, 0.12788299057773408, -0.4526766173311968, -0.1882013624076047, 0.2613861914994503, 1.1664718536555214, 1.110828842770521, 0.023993327994260932], "ratio96": 0.9999999999727042, "ratio_final": 0.9999999999797999, "diag": {"ratio": 0.9999999999797999, "nonnormal": 0.026586097386214255, "eig_gaps": [0.29792812482622527, 0.7528144692765273, 0.9988457557940946], "discness": 0.5328967851020309, "peaks": 1, "deg": 3, "diamW": 4.458796762420723}} +{"family": "companion", "seed": 604007, "x": [-0.7437880378231271, 1.0143845499147166, -0.8243501636679489, -0.8556911519520369, -1.1760441395784644, -0.6864533086341797, 1.2168401818021288, -0.1296392136775943, 0.6707746122276355, -0.3255476336974368, 1.1439339432255933, 0.03397712589622506, -0.8074943892469328, -0.25307581691281766, -0.12511933452404866, -0.3314547812205869, 0.5631564589687239, 0.12430382115701638, 0.6930445827479801, 0.09554675134526147], "ratio96": 1.6497401469219524, "ratio_final": 1.6494641836904647, "diag": {"ratio": 1.6494641836904647, "nonnormal": 0.5853250747270698, "eig_gaps": [0.1421379117609941, 0.6391030697693364, 0.6816773654077716], "discness": 0.4487763205000882, "peaks": 4, "deg": 3, "diamW": 2.745888722199415}} +{"family": "peaks", "seed": 605007, "x": [-0.1765898024295985, 0.3973552834877251, 0.031397974038863394, -0.3680816797706945, -0.9447651964782122, -0.6517984832669272, 4.211456028591624, -4.300165444682208, -0.6071026710009543, -0.3315498298007954, 5.937619270454003, -0.4255896950705138, -0.10074921238966414, 0.48470500207053185, 2.5086271156891238, 1.85145439104362, 3.3796711949965075, 0.08236626968640567, -0.026033216815534145, 0.07051343562158108], "ratio96": 1.9983515309414068, "ratio_final": 1.9983602386976598, "diag": {"ratio": 1.9983602386976598, "nonnormal": 0.7115370492397998, "eig_gaps": [0.0924183005062884, 0.1184437192826544, 0.1515054550959211], "discness": 0.002737086470106558, "peaks": 3, "deg": 3, "diamW": 8.582623087335197}} +{"family": "real", "seed": 606007, "x": [-0.6151886362426193, -0.08152155331772727, -0.6606707545365024, 0.1317814011820978, -0.06123834051670721, 0.15066678146271056, -0.06774711704762004, -0.07914521744676505, -0.5354278445369385, 0.25945135223097765, 1.8382864684430495, 0.25997265467715924, -0.6701710879328178, 0.29518397824760245, -1.9243637927153379, 0.02214011678336568, 0.030793185360431838, -0.017331491726974152, 0.04390353328513642, -0.001342677952164142], "ratio96": 1.9958947807005225, "ratio_final": 1.9963184216629908, "diag": {"ratio": 1.9963184216629908, "nonnormal": 1.2023536259774834, "eig_gaps": [0.10686893498297914, 0.29387002500016596, 0.29431761734762635], "discness": 0.02623759878684691, "peaks": 4, "deg": 3, "diamW": 2.0407996232833407}} +{"family": "random", "seed": 600008, "x": [0.5482634954677115, -0.299567343125135, 0.8339297211755294, 0.6846719204862453, 2.002667724099597, 0.4695714410048557, -1.2153104003004334, 0.6787635599404432, -1.7382846966071155, 0.23154728149793855, -0.734200679651765, 0.03874989226563717, 1.0575939649486854, 0.5208838254262913, -1.1000389030809221, -0.22719606738642104, -0.10673889806581191, 0.17549066376895378, 0.03973983992577486, -0.042670478937061135], "ratio96": 1.8890659814715745, "ratio_final": 1.8890855571076424, "diag": {"ratio": 1.8890855571076424, "nonnormal": 0.822547496930886, "eig_gaps": [0.3595429669904677, 0.41690600573691644, 0.5771927447176084], "discness": 0.21483406065415805, "peaks": 3, "deg": 3, "diamW": 2.850444080698576}} +{"family": "jordan3", "seed": 601008, "x": [0.3144187325665587, -0.18575203670939486, -0.21504335122552581, -0.10853784669968543, -0.43052648678479544, 0.08600549270520569, 1.2732475711446907, 0.17279164951200213, 0.24460456204068093, -0.19125320291892253, 1.0833282424002353, -0.5822306288580158, -0.06766390234332091, 0.09639171599360932, 0.3572780724956082, 0.22675233234629627, 1.5053521403353063, 0.15566351206502066, -0.226102968952146, -0.006682338925734668], "ratio96": 1.988205468037788, "ratio_final": 1.9881820015402822, "diag": {"ratio": 1.9881820015402822, "nonnormal": 0.7227059729623553, "eig_gaps": [0.14732621988944067, 0.2715327687653453, 0.40241324644991006], "discness": 0.03340913786013266, "peaks": 3, "deg": 3, "diamW": 1.9705273011058309}} +{"family": "jordan2", "seed": 602008, "x": [-0.28405817948517365, 0.1806493940053815, -0.3944030929736102, 0.1866288220882787, -0.36314847976788145, 0.20587910802983544, 0.3543830947735134, 0.13520670188055783, -0.041968852759149994, 0.0273578958061311, -0.015532622208690394, 0.009151275599596543, 0.5327853667845723, -0.1214754465022394, 0.8512062560561802, 0.3144321463322617, -0.5675158180895041, -0.31720865100516726, 0.3734599255196807, 0.6905925565015956], "ratio96": 1.9372081271870614, "ratio_final": 1.9372096446727198, "diag": {"ratio": 1.9372096446727198, "nonnormal": 0.34381499487903494, "eig_gaps": [0.09215199820703723, 0.20841037563267303, 0.27742243207001877], "discness": 2680711282.6397796, "peaks": 2, "deg": 3, "diamW": 0.3983340572052407}} +{"family": "nearnormal", "seed": 603008, "x": [0.4406406908494128, 0.9952198993497483, 1.30036157648349, -0.6470946330921177, -0.7481852834351249, -1.6493604991910074, -0.834272157715913, 1.3998340575693882, 1.6183664322634361, -5.694141860204088, -1.1810542330392777, -0.5308794939238032, 0.12510015552246123, -0.5076516774084071, -1.2755877873408201, 7.043141987809309, 0.1486449499869461, 0.025131633622950747, -0.02158999908884418, 0.003116592568198883], "ratio96": 1.9667115082177578, "ratio_final": 1.9670049320670957, "diag": {"ratio": 1.9670049320670957, "nonnormal": 1.340063315758835, "eig_gaps": [0.26765939925503784, 0.3292931526298168, 0.41865792939732044], "discness": 0.03081778349759919, "peaks": 4, "deg": 3, "diamW": 6.925706979684686}} +{"family": "companion", "seed": 604008, "x": [-0.2754563175082258, 0.17265408307331592, 1.3536145022884947, 0.8097190008367943, 1.309742074114017, 0.8233509571289479, -0.0012833211786059755, 0.024160880543536163, -0.020601971227375247, 0.007771326074972624, 1.2021600453532069, 0.009956727647255451, 0.11351582909117386, -0.6837486465075397, -0.2161108200227273, -0.26914960406275357, 0.46074031992624387, 0.10020890047051029, -7.127256550480243e-05, 0.09217994967042087], "ratio96": 1.7672376458983716, "ratio_final": 1.622248131205978, "diag": {"ratio": 1.622248131205978, "nonnormal": 0.3182435267271336, "eig_gaps": [0.019692726214498957, 0.7345106217806954, 0.7497938716187902], "discness": 229741440081.4133, "peaks": 1, "deg": 3, "diamW": 2.3329165321236696}} +{"family": "peaks", "seed": 605008, "x": [-0.7477112935789108, -0.28358036524013674, -0.4678700580987505, -0.21884097137186664, 0.9261623693922111, 0.17081245531790867, 1.1892311176438057, 0.0001253994736189667, -0.018636355851219838, -0.011159107011227073, 0.018325474671064525, -0.016141605181029212, -0.6574223280087002, -1.8152894798197725, -1.2602298372620813, -0.9922060938248987, 1.4962512061315874, 1.4662287094996533, 0.5882373003137391, 0.12182396713467504], "ratio96": 1.8380817392523627, "ratio_final": 1.744658482946865, "diag": {"ratio": 1.744658482946865, "nonnormal": 0.641763085196776, "eig_gaps": [0.1303888057630317, 0.6570757912943254, 0.7873533921528832], "discness": 0.3484063465543013, "peaks": 1, "deg": 3, "diamW": 2.202889620666029}} +{"family": "real", "seed": 606008, "x": [-2.622064035328334, 0.0571702830214681, 0.542352880121455, -0.01413354701221095, 0.7690079284825266, -0.03242538330513289, 8.435894367132835e-06, -1.4193304733438171e-05, -1.2779171572293208e-05, -8.64223751138383e-06, 0.4881015858949144, -0.11579077576896114, 0.46050241676982806, 0.20003761566211614, 0.9156377744261623, -0.1189037599114616, 0.6310289659164243, -0.0799183730878677, -0.8139797690448387, -0.13142543325965544], "ratio96": 0.999999999965622, "ratio_final": 0.9999999999771068, "diag": {"ratio": 0.9999999999771068, "nonnormal": 0.048739896951026235, "eig_gaps": [0.06398110411835486, 0.8905956171817622, 0.9544763420564613], "discness": 1.3841769024292512, "peaks": 1, "deg": 3, "diamW": 3.554048658343728}} +{"family": "random", "seed": 600009, "x": [-0.2164169344209485, -0.019747824112128887, -0.3008146097829501, -0.41085406844582595, -0.3571856208042723, -0.01670345958414328, 0.5305050817572354, -0.22080049178383426, 0.040928195294475694, 0.8571630482677889, 0.5509943936530065, 0.4223772041132204, 1.5565260931510856, 0.1821270504997255, 2.8850108727434534, -2.6685427496953396, 1.1192857391848252, 0.4762969663055831, 0.302847829450731, 0.09892541670860383], "ratio96": 1.9442929439253747, "ratio_final": 1.9457992572343816, "diag": {"ratio": 1.9457992572343816, "nonnormal": 1.1435837574526493, "eig_gaps": [0.1082457621860932, 0.30609928396757774, 0.30759652350093303], "discness": 0.03445977897098319, "peaks": 3, "deg": 3, "diamW": 1.300758565474143}} +{"family": "jordan3", "seed": 601009, "x": [-0.15452597856081968, -0.2839683438384515, 0.09300237593242727, 0.09847117048060361, -0.03996682658513602, -0.07811844713263572, 0.888890671325858, 0.47327520700076803, -0.1558962019492356, -0.0748215395879191, 0.8034979112159288, 0.6845926427845052, 0.00047005054978060197, -0.006927384464892248, 0.07800210973077021, 0.2622009402706389, 1.0474364152827778, 0.20555117948146984, -0.11728884255402644, 0.12751776275518656], "ratio96": 1.9930046020093364, "ratio_final": 1.9930006270504441, "diag": {"ratio": 1.9930006270504441, "nonnormal": 0.7024397595631039, "eig_gaps": [0.14378460892338557, 0.15323337214627833, 0.2963162696688254], "discness": 0.018006644699606302, "peaks": 3, "deg": 3, "diamW": 1.5373941570266718}} +{"family": "jordan2", "seed": 602009, "x": [-0.19452431237261925, 0.09804837037695308, -0.008373245083055543, 0.2984988326192781, -0.309551064039311, 0.2748523537743246, 0.20590487262715007, 0.2790996164689349, 0.5365513785429842, -0.18216361816799043, 0.6701366491913361, -0.1831037259945752, 0.08436287406992354, -0.2215751657254243, 0.6373239593855158, -0.002993022550323708, -0.06283572381758612, 0.0026877576193868307, -0.11423064119124599, -0.030049259237089287], "ratio96": 1.9691040387895242, "ratio_final": 1.9692063347728552, "diag": {"ratio": 1.9692063347728552, "nonnormal": 1.0711010143670976, "eig_gaps": [0.2099987158342068, 0.27234973725392303, 0.30077308254592383], "discness": 0.07522100197009585, "peaks": 4, "deg": 3, "diamW": 1.0044272324409476}} +{"family": "nearnormal", "seed": 603009, "x": [-0.14125951240596646, -1.045088897498958, 0.21486123972053234, -1.1963393869777597, -0.24984627733173906, 2.0516673950431725, -0.4115836268199198, -0.2220817349609956, 9.182810360824608e-06, 6.78951835647619e-06, 5.45644495742951e-06, -1.4128577640834595e-05, -0.052495175927409585, 0.0066326006355122815, -0.1385833654375104, 0.268051571467903, 0.38430424135099617, 1.0186433194489815, 0.9643854846529082, 0.13099834458989384], "ratio96": 0.9999999999705363, "ratio_final": 0.9999999999813017, "diag": {"ratio": 0.9999999999813017, "nonnormal": 0.05670174667421663, "eig_gaps": [0.11257282322987744, 0.9015678092679277, 0.9546444851364765], "discness": 0.5686714704213492, "peaks": 1, "deg": 3, "diamW": 3.4369677531509044}} +{"family": "companion", "seed": 604009, "x": [1.2922586602296087, 2.7707623149437524, 2.9125009240157507, -0.37938020237817804, -3.113706695687682, 1.545549517831895, 2.475753327274261, -0.9667452291793286, -2.1884973533737884, 0.7152035096887757, 3.156829427079924, 0.10014328876174808, -10.94925612077938, 1.8758524019763825, 2.0856689459588265, -6.057786270380361, 3.2893842322686746, 1.48974287416753, 0.1226484233609037, -0.0037053112865519755], "ratio96": 1.638834562593091, "ratio_final": 1.6488330511479956, "diag": {"ratio": 1.6488330511479956, "nonnormal": 0.6534134011379263, "eig_gaps": [0.4494054594797422, 0.5801712472833456, 0.8025691003900688], "discness": 0.1906597734634174, "peaks": 3, "deg": 3, "diamW": 7.882409579119808}} +{"family": "peaks", "seed": 605009, "x": [-0.9770454118459148, 0.5789408300213066, -0.1661107948295532, 0.08098090923521865, -0.009381243077366484, 0.5770246055507922, 1.2312652198812881, -0.08006190456450783, -0.31406852558517645, -0.03930833011110445, 0.11024209482619048, 0.0886714697952103, -0.20501751380853106, 1.9422176938266977, -1.3004541738138444, 1.8852716626047639, 1.3317971907853434, -0.12113918848090645, 0.5958310778248777, 0.1889152992556235], "ratio96": 1.9213514804747929, "ratio_final": 1.923676024962603, "diag": {"ratio": 1.923676024962603, "nonnormal": 0.8649163526050839, "eig_gaps": [0.3269720448734276, 0.5981237524528749, 0.6082097541344051], "discness": 0.9564202800672704, "peaks": 5, "deg": 3, "diamW": 1.5910071476131389}} +{"family": "real", "seed": 606009, "x": [-0.5689875963800747, -0.18222037832968788, 1.0323002692075458, -0.456675383666695, 1.4205340104631095, -0.42657549581237353, 0.00941851681748685, -8.756801144055837e-05, -0.0039375193445496965, 0.008208539000465599, -1.338891670585358, -4.451387435844728e-05, 3.324988887650994, -0.841958699280823, -0.22777631602393977, -0.7560910115971269, -2.2654118649659747, -0.18837802480073096, 0.40368039163885605, 0.10408042294737152], "ratio96": 1.8785476750536976, "ratio_final": 1.6579877364104212, "diag": {"ratio": 1.6579877364104212, "nonnormal": 0.46951892266010653, "eig_gaps": [0.15513918229279236, 0.6472669812239553, 0.7985952703625642], "discness": 1.9861807191220506, "peaks": 1, "deg": 3, "diamW": 2.5099966001417213}} +{"family": "random", "seed": 600010, "x": [-0.9647657048168763, 0.1392075826832031, 0.7627283799601545, 0.6685157873257976, 0.5466804302019641, -1.2198553410155584, -2.600587905310131, 1.4474814654757742, 1.276208665006937, -0.941381438603377, 2.2694837775804797, 2.1260996088889943, -0.9561468614661242, -0.8130358606492252, -1.4981896804376529, -1.3456689135179758, -1.0719616363257494, 9.286292965470945, 0.5823426512417014, 0.2239526101372098], "ratio96": 1.9429627572578805, "ratio_final": 1.943184189433361, "diag": {"ratio": 1.943184189433361, "nonnormal": 0.7411167999888572, "eig_gaps": [0.3600297208724793, 0.3787457303467211, 0.40503399563422265], "discness": 0.012665356300774084, "peaks": 3, "deg": 3, "diamW": 5.018379837105981}} +{"family": "jordan3", "seed": 601010, "x": [0.5904444312723007, -0.08024797934183736, -0.12132562466026747, -0.16176682695999922, -0.4647740918811282, -0.15554204454529585, 1.1135816378252572, -0.08115981571009254, 0.04586179742212153, -0.09520779972082072, 1.087353356769134, -0.5716570074524002, -0.26791728337751375, -0.030865864613020877, -0.08042073664650118, 0.4003510776859256, 1.7391294756385638, -0.03913508183805342, 0.10846030908972132, -0.060498110273579486], "ratio96": 1.977725341824112, "ratio_final": 1.977700890059782, "diag": {"ratio": 1.977700890059782, "nonnormal": 0.6877018385727185, "eig_gaps": [0.17322457083249462, 0.3612818303676745, 0.533484462524563], "discness": 0.07194895273053578, "peaks": 4, "deg": 3, "diamW": 1.9830031676546678}} +{"family": "jordan2", "seed": 602010, "x": [0.27929163963725523, -1.081113637630387, 0.17306812483781364, -0.48256335045599874, 0.43218601227519327, -0.47993174971118924, 1.1675685630610204, -0.09297288864977585, 0.6468213314331627, 0.2563949110376399, -0.12964184097222298, -0.14417053595525875, -0.23553825728809508, 1.5806588224409692, 1.6082872591610675, -0.2266093231821642, 0.13102842182725138, 0.48448155763407885, 0.1524924640021304, -0.09897354790871543], "ratio96": 1.9828324112171216, "ratio_final": 1.9829410234617277, "diag": {"ratio": 1.9829410234617277, "nonnormal": 0.7553613045645912, "eig_gaps": [0.1729015570179441, 0.40561433650324774, 0.41389919946457515], "discness": 22710183687.991165, "peaks": 4, "deg": 3, "diamW": 1.4987213234399015}} +{"family": "nearnormal", "seed": 603010, "x": [-0.6093176821268349, -0.28684300635235593, 1.1555752897359537, 0.93970200953181, 0.6863889043002778, 1.8258550889360414, 0.9351933438147279, 0.33388682250426444, 0.10157652740274954, 0.5759301060623156, -1.6256294283431778, -1.0586700613167288, -1.350220452584944, 0.7937964281321792, -4.9470227412257195, -0.8004919098319574, 1.9901669261720123, -3.4585486486898542, 0.37253431117680447, 0.3889819682781135], "ratio96": 1.8265228027520146, "ratio_final": 1.8307680805342925, "diag": {"ratio": 1.8307680805342925, "nonnormal": 0.5616452305243439, "eig_gaps": [0.2890270807613838, 0.6195189781591827, 0.7143904729507815], "discness": 0.4753444182842701, "peaks": 3, "deg": 3, "diamW": 3.4692179367615994}} +{"family": "companion", "seed": 604010, "x": [0.32541938807110826, -0.7896395486610123, 1.128309187652112, 0.3712361326262055, 2.6916881978551856, -0.5049974516483481, 1.511355672227492, 0.3925218101766922, -0.09999899257959445, 0.07999440556052599, 0.2100223434622077, 0.20996386340981926, -0.2975914393258991, 0.3182634687769481, -0.13135990844478299, 0.09033756296046282, 0.7068648534278181, -0.05844877228577516, 0.9746660911347274, -0.03782453818242493], "ratio96": 0.9999999999740526, "ratio_final": 1.0027249091450632, "diag": {"ratio": 1.0027249091450632, "nonnormal": 0.3815171986764045, "eig_gaps": [0.49267072732404044, 0.6255570652200253, 0.8318920400010905], "discness": 194814427293.44205, "peaks": 1, "deg": 3, "diamW": 2.864947882815614}} +{"family": "peaks", "seed": 605010, "x": [-0.9320539125737016, 0.14053406042985958, -0.23030552286095932, -0.02481341023777415, 0.7058953164050452, -0.8931076858592911, 1.9336263005424366, -0.030903682458190362, 0.0033064041077288447, 0.44123017016606536, 1.6941423955083896, -0.285932050939178, -1.3943829669014896, 1.685723066873059, 0.22819379759063463, 2.6934558303377414, 3.568422137914287, 1.100451344487354, 0.1768570568214559, 0.10487056835788253], "ratio96": 1.9538452530896613, "ratio_final": 1.954813825118366, "diag": {"ratio": 1.954813825118366, "nonnormal": 0.6910917299827187, "eig_gaps": [0.21989912896923347, 0.38945502664403614, 0.590744644108898], "discness": 0.10321811476562459, "peaks": 4, "deg": 3, "diamW": 3.278617078270614}} +{"family": "real", "seed": 606010, "x": [-0.6686592052778355, 0.03385692084974145, 1.1094219045118678, 0.33570108235098317, 0.7053445338321562, -0.2438736725531393, -3.1503493680630386, -0.5231458638000837, -2.3703533659846805, 0.09060084619947861, 0.06316201088904805, 0.2955441535848159, 0.22392191301683706, 0.22784969215175627, -2.6950458530766848, 0.32505489340198657, 0.034664744957582594, -0.04976159538227004, 0.022215673384456597, -0.01165184371373916], "ratio96": 1.980061967603888, "ratio_final": 1.9801538184736822, "diag": {"ratio": 1.9801538184736822, "nonnormal": 1.3386684041387473, "eig_gaps": [0.16242788027490174, 0.3222650293962243, 0.4146201999450412], "discness": 0.03330029299797145, "peaks": 3, "deg": 3, "diamW": 4.349810813006577}} +{"family": "random", "seed": 600011, "x": [0.8009286265553185, -0.35446897951504064, 0.07440495143561181, -1.1365597863955927, -0.7836667242539859, -0.25521229230899856, 0.3304005056466759, 1.886074756051021, 0.08528501463876817, 0.9625671743430814, -0.8601619564969181, 1.6475845176709862, 0.013366787779031976, -1.59936676717848, -3.17962889185179, -0.13690403607100038, -0.14383960830048062, 2.431722886943943, -0.3280384002869576, 0.06166203197400973], "ratio96": 1.9124954869643978, "ratio_final": 1.91721205478952, "diag": {"ratio": 1.91721205478952, "nonnormal": 0.6897229123400266, "eig_gaps": [0.3287242908534352, 0.378793270523012, 0.4889256558302207], "discness": 0.04511909418720725, "peaks": 3, "deg": 3, "diamW": 3.2473259153877763}} +{"family": "jordan3", "seed": 601011, "x": [-0.1403988355693968, 0.361980362883084, -0.10796959397809319, 0.28981575938266046, -0.0730112933009956, 0.4271948775030475, 0.24373332679315873, 0.2560565375883579, 0.023267637084623406, 0.10700316143982164, 0.011287871926688356, -0.03550339951293588, 0.18104507325772745, 0.4749574765019463, -0.6993261563487687, 0.6552296456085518, 0.9153196310220455, 0.4901782388929077, 0.4571817573865397, -0.019930979351225814], "ratio96": 1.969419348853536, "ratio_final": 1.9694424988248191, "diag": {"ratio": 1.9694424988248191, "nonnormal": 0.35039223360605026, "eig_gaps": [0.2075952312785431, 0.24606230427486614, 0.3719603559484212], "discness": 1435906517.4969554, "peaks": 3, "deg": 3, "diamW": 0.38110836737956805}} diff --git a/problems/crouzeix/data/census_w1.jsonl b/problems/crouzeix/data/census_w1.jsonl new file mode 100644 index 0000000..69a6543 --- /dev/null +++ b/problems/crouzeix/data/census_w1.jsonl @@ -0,0 +1,19 @@ +{"family": "random", "seed": 100000, "x": [-0.4693482384723695, 0.5606797722517263, 1.22528817159415, -0.43567083789954375, -0.6795273738738636, -0.30378262430531733, -0.6297257982192327, -3.5249875132345547, 0.18851235213902695, -0.9064018304154038, -3.448427192307559, 0.28016577576209023, -1.3899388091492442, 0.826307405356063, 1.0374241682979481, 0.08188603174201664, 4.783392829889163, 0.9044251590751338, -0.25973072926373575, -0.11733388971453573], "ratio96": 1.9730777171334675, "ratio_final": 1.9730790509888327, "diag": {"ratio": 1.9730790509888327, "nonnormal": 0.7137613572322133, "eig_gaps": [0.16159029350180465, 0.3468082060913677, 0.3570630886043997], "discness": 0.025587354690489943, "peaks": 3, "deg": 3, "diamW": 5.505567574650626}} +{"family": "random", "seed": 100001, "x": [0.5203954203589413, 0.4186912687683316, 0.782607554009086, -0.6918064853130799, 0.3618122405963487, -0.6530564709920417, 1.1680238904544085, 1.5904475709165093, 0.9070337387321544, 0.18629165326981678, -0.2011676894440021, 1.8896963315180075, -0.22291182448109179, -1.8574186717159689, -0.7777280120589901, 3.678208044467908, 1.6764566763704514, -1.9131281040921406, 0.4273480301605477, 0.004481990210881113], "ratio96": 1.9453020924988285, "ratio_final": 1.9454260507662304, "diag": {"ratio": 1.9454260507662304, "nonnormal": 0.684407907980499, "eig_gaps": [0.13459303878812942, 0.3450750640888819, 0.3634267849831733], "discness": 0.04264622441053452, "peaks": 3, "deg": 3, "diamW": 3.139655273293216}} +{"family": "random", "seed": 100002, "x": [-0.10761593427850549, -1.1825688711613145, -0.12947956518585124, -1.3593847517763082, -0.04927260230867228, -1.0821474541347444, 0.6149112783994763, 0.07992189594406728, -0.06863344608840112, 0.21180176695142008, 0.12964171232901778, 0.015634019224478768, 0.7182466927553433, 2.35759862560332, 0.5920982415051725, -0.6802454290243981, -0.4323296167326621, 0.7183647369494864, -0.1463951815967059, 0.0774973592275475], "ratio96": 1.9658337699795976, "ratio_final": 1.9658338189373585, "diag": {"ratio": 1.9658338189373585, "nonnormal": 0.1337537103673971, "eig_gaps": [0.16789234009972617, 0.2575531459870722, 0.41721180889533344], "discness": 3437866816.164902, "peaks": 3, "deg": 3, "diamW": 0.6917504057179872}} +{"family": "random", "seed": 100003, "x": [-8.539297932575458, 4.662726037460518, 0.07273790925563822, 1.028118900236058, 0.8090269418860376, -1.0258483067960487, -4.706364470695651e-06, -1.99399966615987e-06, 1.938701010736351e-06, -1.3557192736694334e-06, 1.6560700360789555, -0.08602661168368617, -0.7991828140297721, 0.548197300771383, -0.9107805926093497, -0.33098929554129014, 1.058014039495844, 0.07710195359262512, -1.1590147625157772, -0.37939946345629766], "ratio96": 0.9999999999653804, "ratio_final": 0.9999999999999278, "diag": {"ratio": 0.9999999999999278, "nonnormal": 0.0641569723312452, "eig_gaps": [0.1933872324441422, 0.8284818956729327, 0.9698907458808456], "discness": 1544420728328.4534, "peaks": 10, "deg": 3, "diamW": 11.28279923451957}} +{"family": "random", "seed": 100004, "x": [0.3961471874611877, -1.649764809556161, 0.1672914997542405, -1.0533666284966348, -0.11917853536206535, -0.6382581813903143, -1.0685708267423633, -0.4861571894849772, 0.1375283021699676, 1.0367410569747728, -0.1307395501674027, -0.08212472736410398, -3.7101178302324413, 3.0384205362847645, -0.11469673241089628, 1.388820354855394, 0.20378629808110682, 1.9260164713962622, 0.08590739485707609, -0.23541741453467846], "ratio96": 1.7574423362270066, "ratio_final": 1.8343911617633484, "diag": {"ratio": 1.8343911617633484, "nonnormal": 0.5804259291457785, "eig_gaps": [0.26797355845505916, 0.33940280529035566, 0.603152791494703], "discness": 54277044991.53609, "peaks": 3, "deg": 3, "diamW": 1.8821300419546991}} +{"family": "random", "seed": 100005, "x": [-1.4935464611800648, -1.73059755527555, 1.6988864698098056, 0.8565570954112128, 1.4315053066127517, 0.7390262001374228, 0.1499891499775842, 0.15001988462705437, 0.16001175879908608, 0.1600181422866594, 1.0093312375145804, 0.6182686261929704, 0.5345175940123149, -0.7286601896235778, -1.0616364743443532, -0.06557360364574379, -0.11204433994459281, 0.2961641202380269, 2.1046944054042216, -0.8165585901484685], "ratio96": 0.9999999999653757, "ratio_final": 1.0613982600645762, "diag": {"ratio": 1.0613982600645762, "nonnormal": 0.22342849408437326, "eig_gaps": [0.06367191502664943, 0.8345456092290574, 0.8957939315324392], "discness": 0.4957661357273988, "peaks": 2, "deg": 3, "diamW": 4.587144071392523}} +{"family": "random", "seed": 100006, "x": [0.866855044325146, -0.8782257982727499, 0.8694654488006663, 0.7438511532544664, 0.7728125455024923, -0.5580193472333204, -0.0835897101312655, -0.02396449423394648, -0.34269307607572724, 0.655384484125632, -0.051425223061235655, 0.08221397858010786, 1.1672458472685376, -0.9064223272604035, -0.35907704609346, 0.0795080231557613, -0.8271017330839479, 0.5532455992277013, 0.9367687891956982, -0.4495051181781817], "ratio96": 1.7285252713844548, "ratio_final": 1.6810540159458285, "diag": {"ratio": 1.6810540159458285, "nonnormal": 0.2079840914399782, "eig_gaps": [0.17850449501027432, 0.698255564318281, 0.8676109943271771], "discness": 209493532389.32172, "peaks": 1, "deg": 3, "diamW": 1.8695925507981073}} +{"family": "random", "seed": 100007, "x": [1.408938420047168, 1.1049439721822487, 1.555098795046052, 1.2206975734502825, -0.4106576426173465, -1.0390088503111072, 0.30809928785882, -0.7919916947534265, -0.0769250078170034, 0.1504114454596065, 0.14825345234683962, -0.0351241121603172, 1.3473526681130714, 0.2329311015034383, 0.5258260119304556, -1.2883326464709006, 0.7811150853674307, -0.15559983368036057, 0.06863208464556098, 0.6948168958916812], "ratio96": 1.7695600072957667, "ratio_final": 1.5277950666639002, "diag": {"ratio": 1.5277950666639002, "nonnormal": 0.15122621461439986, "eig_gaps": [0.05575240106039937, 0.8408748239788189, 0.895612994251482], "discness": 1.076159926011765, "peaks": 1, "deg": 3, "diamW": 3.3441615498697104}} +{"family": "random", "seed": 100008, "x": [-0.22535346106510534, -0.3246253418637256, -0.06354535724912982, 0.21343218316144996, -0.04033496507438562, -0.17374405756708577, -0.31972801764933667, -0.27348458130534875, -0.16866309483830194, -0.5160743851628061, -0.2848273856988516, 1.063846425164565, 0.1527454242650909, 0.2986503834268165, 0.2823036635391613, 2.669202164003307, -0.004301594057581123, -0.3690912972196131, 0.05369520437051117, 0.11686180572873747], "ratio96": 1.96941740382222, "ratio_final": 1.973967535275769, "diag": {"ratio": 1.973967535275769, "nonnormal": 1.256209215644697, "eig_gaps": [0.1790460479424658, 0.29088839184203125, 0.42137386147293165], "discness": 0.010760797276092134, "peaks": 4, "deg": 3, "diamW": 1.3334025506054474}} +{"family": "random", "seed": 100009, "x": [0.883549468394631, 0.7114839557841446, 0.3691278423747475, -1.048255324895143, -1.2058649234395178, 0.5054331497850711, 0.9658162013352815, -2.5089092841802305, 0.26163278564103043, -0.14782920263576088, -2.7833631324519548, -0.7666643893497079, 1.0055702398309105, -1.4460783754080409, 0.37140143049063756, 1.010145395127331, -2.130413012416887, 2.066669388453758, 0.3155536265255653, 0.2679355564425362], "ratio96": 1.9417827653610666, "ratio_final": 1.942288718514082, "diag": {"ratio": 1.942288718514082, "nonnormal": 0.7756922101391138, "eig_gaps": [0.40225826317346236, 0.4606560173733287, 0.48540802644036146], "discness": 0.02669466194273573, "peaks": 4, "deg": 3, "diamW": 4.5577388292963885}} +{"family": "random", "seed": 100010, "x": [-1.5913379424991982, -0.369571518327837, -1.939944648634464, 1.5684761264231193, 0.4913866740332543, 1.7224500836102825, -1.625809391872662, -0.12771444289756198, -1.5313319631942086, -0.4053343557704905, 1.4245804918950022, -0.6921851399158407, 0.22872661691290896, 2.6828113442777153, -1.1997067174971006, 0.9781816671284244, 0.6782795893563587, 1.764677420391465, -0.3031157769865825, 1.2889929638141282], "ratio96": 1.479349774237704, "ratio_final": 1.4999138730193975, "diag": {"ratio": 1.4999138730193975, "nonnormal": 0.5493095038990424, "eig_gaps": [0.48457054312775727, 0.5995029124318217, 0.7264306642639918], "discness": 0.23201055709987173, "peaks": 3, "deg": 3, "diamW": 4.063703290614798}} +{"family": "random", "seed": 100011, "x": [1.4626496829419486, 1.2946046115497407, -1.5554000634917475, 1.031956373329204, -1.3201002791530043, 0.8891186942290337, -1.5085136068958895, -5.746169054539017, 2.444995743724343, 0.4763911508213017, 6.9150306407196585, -1.5010913467717295, -0.8620147671516013, 3.4253782151811087, -1.0861512725567382, -2.394418890504598, 0.565866488278207, -0.9389187548185367, 0.012928107826576954, -0.10957757537484092], "ratio96": 1.7911731889735683, "ratio_final": 1.8080969842886498, "diag": {"ratio": 1.8080969842886498, "nonnormal": 0.8708825700618938, "eig_gaps": [0.02768309762312823, 0.28281770839722503, 0.3046736104705863], "discness": 0.029892918580847104, "peaks": 3, "deg": 3, "diamW": 9.943285777825906}} +{"family": "random", "seed": 100012, "x": [0.8967272567614075, -0.15651867014372695, 0.5398891923780852, -0.4674217709579496, 0.8970601297828384, 1.3727485965576989, 0.03161356890473382, 0.4494922266254775, 2.8951017641286914, -0.061818973106778495, 2.0829344418107327, 0.26060728266378985, -7.745218066958293, -2.421821808767242, 7.127150618046123, -0.6845853931579513, 0.7077294632141942, 0.45945784020154345, -0.018871110219946495, 0.14342348765078133], "ratio96": 1.8738062500904267, "ratio_final": 1.8764272748780273, "diag": {"ratio": 1.8764272748780273, "nonnormal": 1.144760971694565, "eig_gaps": [0.1224074748229131, 0.3955241549234103, 0.4848171872777912], "discness": 0.053412524518147195, "peaks": 3, "deg": 3, "diamW": 3.86643213541663}} +{"family": "random", "seed": 100013, "x": [0.13175643655830888, 1.0226764179756258, -0.5686335540539952, -0.5808675251261695, -0.605189211417031, -0.05690850740023082, 0.9736262291901138, 0.1641571196346642, 0.5387001937410778, 0.3499826370393766, 0.911355805525442, 0.1566964300237854, 0.43656832158320524, 2.13018831540922, 0.3107186725367111, 0.7753757671538739, -0.7388318416861706, 0.7961018570244298, 2.5016637565809687, -0.7578221352200192], "ratio96": 1.597413679610732, "ratio_final": 1.6031759311287956, "diag": {"ratio": 1.6031759311287956, "nonnormal": 0.8178727242284826, "eig_gaps": [0.22415051316697931, 0.5578369079720052, 0.7467641321001369], "discness": 0.22659498685072613, "peaks": 3, "deg": 3, "diamW": 2.343214266034889}} +{"family": "random", "seed": 100014, "x": [1.0468110637672692, 0.8232206356930973, 0.18926094822752476, 0.18444533642676636, 0.10829603061002815, -0.028659306202212083, -1.742005772071236, 0.5086619267262648, -0.07487533795516776, -0.23908815039995523, 0.540040685575835, 1.9688058417241363, -0.18310285646294258, 0.3164219720692881, -1.145195388656929, -2.8733772794930275, 1.959198179791363, 1.3310585528106555, 0.16384897016027816, -0.22401381628602846], "ratio96": 1.9886263962978807, "ratio_final": 1.9885905206198318, "diag": {"ratio": 1.9885905206198318, "nonnormal": 0.645178401216383, "eig_gaps": [0.07449051439711116, 0.3494083111647348, 0.4141626082762303], "discness": 0.060569395857676095, "peaks": 4, "deg": 3, "diamW": 3.060348017051216}} +{"family": "random", "seed": 100015, "x": [-2.1082139756864624, 0.23884001126931353, -0.4114913182764382, 3.1501249018710364, -2.051164793159149, -0.43532548139755367, -1.3123815616312808, 3.0748708867493164, 2.3158677784617687, 1.1114821504581536, -3.1646242526117634, -0.7588138137300793, 0.39845018850002556, 3.927162138568426, -1.6479803642655744, -2.173372165956893, 1.5159437798306712, -1.0780797617645228, 1.1990053965647085, 3.4196562781015762], "ratio96": 1.3006169812587736, "ratio_final": 1.3024986626469275, "diag": {"ratio": 1.3024986626469275, "nonnormal": 0.6141943083065777, "eig_gaps": [0.09924812678090553, 0.4942985997425767, 0.5783457402032008], "discness": 0.19077435839326132, "peaks": 2, "deg": 3, "diamW": 6.817005163037982}} +{"family": "random", "seed": 100016, "x": [0.8855717261374931, -0.986801248822396, -1.5510361006529625, 0.3715518529460575, -0.2780486767213678, -1.893455669923906, 0.2625838417742547, 1.9480854490017474, -0.499204578021452, -0.7720579174309954, -1.0063451362533977, 0.6261918697000821, -2.7597104808422155, -0.09493384257917951, -1.8523188499980514, 0.42961912707207095, -0.062899659703959, -0.0437858810585586, -1.1650734209784832, -0.23681707064583124], "ratio96": 1.5552965852362042, "ratio_final": 1.5670927509184347, "diag": {"ratio": 1.5670927509184347, "nonnormal": 0.6731105266698615, "eig_gaps": [0.38550914501288686, 0.6790128386025932, 0.7290418142088296], "discness": 0.1913532425812262, "peaks": 3, "deg": 3, "diamW": 3.826469196766727}} +{"family": "random", "seed": 100017, "x": [1.0287461261539532, 1.2683852202957018, -0.7323843408820017, -0.5147354085025595, -0.22611277050726758, -0.9169059999715552, -0.7817194926779311, -0.15268296797625647, -0.015486163062248218, -0.013524136640631422, 1.020373671088426, -0.5406192907252994, 2.526359674484584, -1.2141298110432592, -0.2425924844006807, -1.355998069414206, 0.7335048292066964, 1.4702372148552274, -0.4344720437208043, 0.22593552765310443], "ratio96": 1.5900209688306248, "ratio_final": 1.6198671741374246, "diag": {"ratio": 1.6198671741374246, "nonnormal": 0.5418713695916001, "eig_gaps": [0.204527672555345, 0.7927837168440192, 0.7971306440680052], "discness": 0.2561837339125295, "peaks": 4, "deg": 3, "diamW": 3.16128115167602}} +{"family": "random", "seed": 100018, "x": [0.09410923068666305, -1.9233788381140888, -2.1206326917619025, 0.09899525629092926, -0.21772330590101086, 1.9926444110562977, 0.6436765457511338, 1.864426444249358, 1.838681079570482, -0.8916715747512178, -0.06476959643985636, -1.7104379142955608, 2.132619424098193, 1.5014730676408752, -1.0169609297919768, -0.3664483077649636, -1.4408405343754165, -2.0032800971151, 0.7650627789199858, 0.9799465095022653], "ratio96": 1.4601121954285377, "ratio_final": 1.4605379236497829, "diag": {"ratio": 1.4605379236497829, "nonnormal": 0.7475897396241189, "eig_gaps": [0.5223141785713838, 0.5835229016256583, 0.7643165229878228], "discness": 0.11619123055840902, "peaks": 3, "deg": 3, "diamW": 5.139780586079189}} diff --git a/problems/crouzeix/data/census_w2.jsonl b/problems/crouzeix/data/census_w2.jsonl new file mode 100644 index 0000000..9da7a3b --- /dev/null +++ b/problems/crouzeix/data/census_w2.jsonl @@ -0,0 +1,13 @@ +{"family": "random", "seed": 200000, "x": [-0.20727378146166764, 0.3981344437327161, 0.4081820052050273, -0.41123356302179004, -0.0069812020865957745, -0.20839220714428153, 0.8402530183646023, -0.8238515450598278, -0.02145183310759179, -1.976025632772055, 0.013509012797120662, -0.7022535634021403, -1.6675684966086641, 1.2646483783964078, 5.248475503882508, 2.468447473716778, -0.09625274812217019, 1.0379814924673356, -0.09181850346029227, -0.14668164440740286], "ratio96": 1.8908738872478639, "ratio_final": 1.8910074964009838, "diag": {"ratio": 1.8910074964009838, "nonnormal": 1.3921095537372083, "eig_gaps": [0.17682283429182985, 0.24443302256585037, 0.38910390201581835], "discness": 0.029905105434007682, "peaks": 3, "deg": 3, "diamW": 2.6131586871093586}} +{"family": "random", "seed": 200001, "x": [1.728664181558529, -1.2054036884764938, 1.7205880017655877, -0.8723556450198927, -1.3247033629840672, -0.40449332191913223, -1.024754762636727, -0.024259417381703056, 0.2684113252725627, -0.08860505434914652, 0.06802138842657543, -0.4517842015150463, -1.3015791014064089, 1.5929895359876634, -0.10062829265360064, 1.8054427059431677, -0.44014665863955266, 0.24617820370383708, 0.7808880296240163, -0.011255383495726214], "ratio96": 1.6892139251234144, "ratio_final": 1.6447805433383824, "diag": {"ratio": 1.6447805433383824, "nonnormal": 0.2355657977807862, "eig_gaps": [0.0910615262496102, 0.8421610333867753, 0.86283624043961], "discness": 1.1243324443745881, "peaks": 1, "deg": 3, "diamW": 3.6584709658843364}} +{"family": "random", "seed": 200002, "x": [0.7036858667098222, 0.03465367302166064, -0.4441043076149801, -1.6405032742836951, -0.3205106030705511, -2.3065517442296946, 0.8263722882289373, -1.3364650321243459, -0.2974900088067053, 0.3475049541015105, 0.2561116329729205, -1.1581896161216365, 1.813390822686939, 0.45413512852034604, -0.2360362620846292, -2.7658374553876577, 0.9272145473532967, -0.34338163177986836, -0.036529281231942407, -0.9148559084221135], "ratio96": 1.7372067786545988, "ratio_final": 1.7372606665458985, "diag": {"ratio": 1.7372606665458985, "nonnormal": 0.4506736411402936, "eig_gaps": [0.20587203050440897, 0.6171305912322188, 0.7766121374117598], "discness": 0.8860817059819379, "peaks": 4, "deg": 3, "diamW": 3.2904839215408703}} +{"family": "random", "seed": 200003, "x": [-1.3743214302535147, -0.5337599163233602, 1.240589860952416, -0.28500973107990013, 1.3149030853138064, -0.8524495610396476, -3.788085856465466, -1.1083464444316453, 0.7950867625766467, 1.4505444755524026, 2.0579208535857303, 3.0633606709243217, 0.8841071593243347, 0.8667409053480515, 2.052859144615721, -1.3649963107947616, -1.4624078818513562, -0.44160139749351257, -0.1303950925533412, -0.008141492571012272], "ratio96": 1.982701264964469, "ratio_final": 1.9826930963618628, "diag": {"ratio": 1.9826930963618628, "nonnormal": 0.7291400387856981, "eig_gaps": [0.08998973031586566, 0.4130413540587588, 0.4258295455270663], "discness": 0.04965228571313416, "peaks": 3, "deg": 3, "diamW": 6.359450744887105}} +{"family": "random", "seed": 200004, "x": [-0.5819813106715443, 1.3147314709442504, 0.8562643129383087, 0.2422458716907742, -0.09755886589176387, 0.5535611226701924, 1.0579334382058905, 0.1881476703714531, -0.4150674472542538, 0.6290396666315244, 1.472498059657569, -0.67416433365892, -0.6308342626055181, 2.000648650167602, -0.9015354163563793, -1.0722545755085147, -0.25608669623326796, -0.41550373421042647, -0.02809745550027475, -1.3098205031623205], "ratio96": 1.6853439113507511, "ratio_final": 1.688628520465699, "diag": {"ratio": 1.688628520465699, "nonnormal": 0.7133592911991993, "eig_gaps": [0.33138163114402663, 0.3685133744861647, 0.65894547955512], "discness": 0.18859961988492482, "peaks": 4, "deg": 3, "diamW": 2.7226754979491634}} +{"family": "random", "seed": 200005, "x": [-1.7763091973999783, 0.8930667031766439, -2.334031671829977, 2.3514070343183366, -0.9822769146387179, 0.49432987177660664, -0.7810617049881149, 2.136792587419132, -0.34727309162392683, 0.42095261288367514, -0.9598914761406885, 0.5271645160790852, 1.9547639183128176, 2.023682025123879, 0.1927026988072379, -5.420294787756656, 2.2758874965918183, -2.5403579112617356, 0.37424600128967933, 0.00851113006009735], "ratio96": 1.728879357216773, "ratio_final": 1.7400715415585755, "diag": {"ratio": 1.7400715415585755, "nonnormal": 0.4608897941007086, "eig_gaps": [0.28133419218022615, 0.4943703514327459, 0.7272836735359233], "discness": 97491640427.07242, "peaks": 3, "deg": 3, "diamW": 3.158257879250238}} +{"family": "random", "seed": 200006, "x": [-1.1867442619669133, -0.18322713634091506, -0.17252403071202732, -0.12682795707977845, -0.4481802417413013, 0.13938432552596336, -0.600682909918804, -0.22064171820991674, 0.911529320654126, 0.4144319803608643, -2.0189309524311017, -0.09511774676553948, -0.08332757922696742, 1.5138934313302532, 3.39078944724512, 5.131137300616616, -0.32429200864747454, 0.8334407199240683, -0.1259297145123836, 0.08430703507708848], "ratio96": 1.9194946462574012, "ratio_final": 1.9320959989558406, "diag": {"ratio": 1.9320959989558406, "nonnormal": 1.0791410412136497, "eig_gaps": [0.1610758628515134, 0.33876112746063536, 0.4269611232353744], "discness": 0.005369616133957167, "peaks": 3, "deg": 3, "diamW": 2.379109244467514}} +{"family": "random", "seed": 200007, "x": [-1.3188147981527751, -0.8812639275998809, -0.47682886667044033, -0.7957519141715963, 0.9012693611816208, 0.4891760055060247, -1.5479060510493396, 1.3702302710517384, -0.9066277830379411, -0.5405935609681444, -1.6266283414428813, -0.12800885049922822, -0.5898128283670115, -1.333092769365011, 0.6031315501988863, 0.41625572682205586, 1.7472976723965745, -0.26321119439716933, 0.2918951296747627, -0.05612905245416744], "ratio96": 1.8334682426179096, "ratio_final": 1.8404060520565506, "diag": {"ratio": 1.8404060520565506, "nonnormal": 0.7543408155798433, "eig_gaps": [0.21664983758480963, 0.4823378359038694, 0.6678812978389246], "discness": 0.16762283477296416, "peaks": 4, "deg": 3, "diamW": 3.9063822863877036}} +{"family": "random", "seed": 200008, "x": [0.27446160385066654, 0.8904057668834517, -0.4717453437440977, 0.154914762331391, -0.2860223425017522, -0.9835358728376744, -1.1476704882262165, 2.7369158914672465, -0.14575978404939383, -0.44169591859480584, -2.487288116964902, -0.14665294296929457, 0.8380322331654317, -1.2305070370056042, 0.9064111750001056, 0.20346545209579048, 4.163104348176575, -0.5527002698213168, 0.16006754871661522, 0.37994579097642067], "ratio96": 1.9604209069900405, "ratio_final": 1.9606714033995576, "diag": {"ratio": 1.9606714033995576, "nonnormal": 0.7455096492967256, "eig_gaps": [0.24173312916681747, 0.2661324521182717, 0.45127496156685043], "discness": 0.03758914559683087, "peaks": 3, "deg": 3, "diamW": 4.3343089680623255}} +{"family": "random", "seed": 200009, "x": [-0.6812149662906526, 0.7982630795488297, -0.32595521020856955, 0.3092947730989428, 1.081417143146723, -0.02793177287718536, 1.6467791494637676, -0.9714703493902487, -0.14292649189095671, 0.17354338695474059, 0.5527816891396454, 1.5644243956102475, 1.7601709357596986, -1.131367287182092, -0.0021364634194132176, 2.0731740838356525, -2.603630044647645, -0.33291986498414916, -0.3753057709105121, 0.24639490358703703], "ratio96": 1.9623745914281074, "ratio_final": 1.9624579595306595, "diag": {"ratio": 1.9624579595306595, "nonnormal": 0.6905489223217623, "eig_gaps": [0.18639708612803516, 0.44631998355729, 0.6003490265760766], "discness": 0.11000790062073583, "peaks": 4, "deg": 3, "diamW": 3.242540575866021}} +{"family": "random", "seed": 200010, "x": [1.0108937698855707, 1.2754605959847085, -1.4877926648331774, 0.8258449262181704, 0.4061811154268076, -2.8149775251594367, -3.5498989124627167, 8.617568031417054, 3.0640871147093156, 0.3646144030796036, -9.037152760847755, -4.05982137782857, 1.1106012120650721, 0.10329074478410809, -0.35299225678509516, -0.5791008259472236, 8.574878320960586, 1.786105826042435, 0.01277539625381129, -0.216120234741269], "ratio96": 1.9934017869517426, "ratio_final": 1.993412142555695, "diag": {"ratio": 1.993412142555695, "nonnormal": 0.7330638223975042, "eig_gaps": [0.17399610393520115, 0.2812641782691535, 0.28338234765448267], "discness": 0.011380715962464708, "peaks": 4, "deg": 3, "diamW": 14.591224927632792}} +{"family": "random", "seed": 200011, "x": [-0.6465181908533031, 1.2704531693633747, 0.568425549659316, -0.11720233352479711, -0.18391333674163968, 0.9562999476082259, 0.732571608246267, -1.3200729957553032, 0.4328730380444179, -0.3621423736765227, -1.7448159943668244, 0.1514445517180306, 1.3601186478751166, 1.0136675955773704, -3.32865050914726, 1.590611514684781, -0.8679640214174233, -1.3333252270030047, -0.40196095201907767, 0.5053116080229865], "ratio96": 1.9333906189423427, "ratio_final": 1.9347119896752438, "diag": {"ratio": 1.9347119896752438, "nonnormal": 0.6247412824654508, "eig_gaps": [0.18547740078019742, 0.43480543705997765, 0.6117537857494899], "discness": 0.14546964848867736, "peaks": 4, "deg": 3, "diamW": 3.014878487729709}} +{"family": "random", "seed": 200012, "x": [-0.05461819400177076, -0.22824944904616407, 0.18326828997632094, 1.6584016657866982, -0.28491744978610434, -0.862597902152686, -1.128874132292126, 5.0246850090524955, -1.019229291316892, 0.1929738707271077, 1.9045701537244657, -4.446399221588061, 2.1310161988267287, -1.482630830684182, -0.08996511180125644, -0.8436753827518877, 2.645067652480142, 0.38626167951106216, 0.06994300533397627, 0.09572981749401205], "ratio96": 1.9763509872238738, "ratio_final": 1.9772965831464346, "diag": {"ratio": 1.9772965831464346, "nonnormal": 0.7311629885305081, "eig_gaps": [0.08876352533174069, 0.25011386147863823, 0.33725382488277456], "discness": 0.025772635467157683, "peaks": 3, "deg": 3, "diamW": 7.60289506306437}} diff --git a/problems/crouzeix/data/census_w3.jsonl b/problems/crouzeix/data/census_w3.jsonl new file mode 100644 index 0000000..d04def1 --- /dev/null +++ b/problems/crouzeix/data/census_w3.jsonl @@ -0,0 +1,13 @@ +{"family": "random", "seed": 300000, "x": [-0.6813971460903179, -1.2848840813851146, -1.339020672011971, 1.0098866639500517, 0.24676675870780712, 1.787344625904916, 1.1722941958586832, -0.4649873870203385, 1.3566799672867726, 0.32755287630465524, 1.3110922453490503, 0.3841055956892929, 2.740360126135154, -0.8018516067419499, -0.2167722315132442, -1.7584518949630745, 2.3042913404746823, 0.38515598678947843, -1.9213134422030715, 0.9175485779198191], "ratio96": 1.4891877278943595, "ratio_final": 1.4918182509573383, "diag": {"ratio": 1.4918182509573383, "nonnormal": 0.6799578250820644, "eig_gaps": [0.44343655822668104, 0.5993638280416301, 0.8058101413026433], "discness": 0.20887319596784268, "peaks": 3, "deg": 3, "diamW": 3.9827909964259254}} +{"family": "random", "seed": 300001, "x": [1.0937842144431515, 0.83274592168023, 0.7293348557125535, 0.6888527270560572, 0.3847719209480996, -0.13623658618806356, -0.6464885414530894, 0.742791424140099, 0.2922793546260005, -0.46862271676625933, -0.3408720627303237, 0.010990409621225187, -0.5184655551958094, 2.0666411323257137, 0.8419492620226359, 1.0222945129604994, -1.8268681646465046, -1.3874322844060658, 0.15226597199832653, 0.5478260939497093], "ratio96": 1.622862989400532, "ratio_final": 1.6745373486800608, "diag": {"ratio": 1.6745373486800608, "nonnormal": 0.4818309828226202, "eig_gaps": [0.23126988882847985, 0.5277553091638165, 0.7086808519724832], "discness": 101182349291.00082, "peaks": 3, "deg": 3, "diamW": 1.6942423532266246}} +{"family": "random", "seed": 300002, "x": [-1.4108468126436782, -0.8150076598169205, -0.8192461175160368, 0.5817137743986165, 1.6013165744567144, 2.4270080768161773, 8.244595481139303, -1.498946253181685, -4.206762601428409, 0.28715153722992176, 0.26656856232427706, -0.43158278179802134, 0.6996688302900804, 5.257847331916345, -4.235903091129684, 11.829835254447799, 0.4454818256010298, -0.14213806448108812, 0.038596871052335106, -0.005835696806673812], "ratio96": 1.6741084789315808, "ratio_final": 1.7273642851384579, "diag": {"ratio": 1.7273642851384579, "nonnormal": 1.280543382167043, "eig_gaps": [0.150064374713862, 0.3011208692167919, 0.4378088112892793], "discness": 0.032150545960073974, "peaks": 2, "deg": 3, "diamW": 10.107969472524063}} +{"family": "random", "seed": 300003, "x": [0.03463492844153197, -0.025642005586560095, -0.18807761262324754, -0.020173804886019272, -0.24481903925963253, -0.061484985280982056, -0.20834696741826642, 0.27657325240557284, 0.5521993764710049, -1.1553441950862882, -0.19773289777498096, 0.14824189551161726, 0.10075134969284388, 0.0811146132469798, 2.7532404811640165, 0.18128867725305275, 0.2448502780840856, -0.15421183358074514, -0.03460527082251129, 0.0029766691142853454], "ratio96": 1.9884591281052564, "ratio_final": 1.988703121753976, "diag": {"ratio": 1.988703121753976, "nonnormal": 1.3551469540154806, "eig_gaps": [0.051177553663092186, 0.1624422149863272, 0.20543613272658412], "discness": 0.007716772043572913, "peaks": 3, "deg": 3, "diamW": 1.3714394413725468}} +{"family": "random", "seed": 300004, "x": [-0.9829717052081945, -0.9446515711907095, -1.8880326164896515, -1.0061706850062677, -0.560049296396607, -0.10880949480801098, 2.0737302016117325, -0.379141897390643, -0.5652476527927236, -0.45991092606560524, -1.6832404689727336, -0.004035576245346575, 2.158819201115916, -1.1343427865993334, 0.7102886056988673, 2.6344218975092897, 1.5571749809987425, 1.4444478337895412, 0.27646930883014437, -0.1920440744574744], "ratio96": 1.6505309955660237, "ratio_final": 1.650451247896331, "diag": {"ratio": 1.650451247896331, "nonnormal": 0.5390381807968079, "eig_gaps": [0.2846968444155609, 0.29398575862865184, 0.5030008966811926], "discness": 7.110133204962462, "peaks": 3, "deg": 3, "diamW": 3.1863693570846494}} +{"family": "random", "seed": 300005, "x": [1.880185524829294, 1.6315069139863851, -0.8761970175452758, 0.03673729447734848, 0.7453622794804844, -0.11257074978164959, -0.4503130269419089, 0.5375450621622844, 1.0439837128594207, 2.296561355857119, -5.835727307908487, -0.9491255864587901, -0.9103875214934811, -0.4806505588259842, 4.184626895137727, 3.0151362401971507, 0.008048640410676334, 0.022446292802832654, -0.01687589183643868, -0.001468933915153878], "ratio96": 1.9526163224866457, "ratio_final": 1.952715171666236, "diag": {"ratio": 1.952715171666236, "nonnormal": 1.241542540416192, "eig_gaps": [0.2350444409525572, 0.3003373410010333, 0.45964538184527565], "discness": 0.026787099398347676, "peaks": 3, "deg": 3, "diamW": 6.92813107293926}} +{"family": "random", "seed": 300006, "x": [-1.6469368732262222, 0.017983387101466387, 1.0883866771580464, 1.0038260359043667, -0.9318371381350692, -0.9497270411623036, -1.617074771904253, 2.144542185973137, 1.0122689854710174, -1.7617243545165073, 2.4777657458099207, 0.8726333400607929, -2.86577216370038, 0.06473508949316581, -0.06278925185681883, -2.0488445187142386, 0.9424676261145213, -2.7062482271504873, -0.02882380623853216, -0.19466501040099315], "ratio96": 1.9237036356739134, "ratio_final": 1.9236629681671338, "diag": {"ratio": 1.9236629681671338, "nonnormal": 0.6840887789911051, "eig_gaps": [0.22479283718237023, 0.5250166976333388, 0.5431897594008801], "discness": 0.09072079220852611, "peaks": 3, "deg": 3, "diamW": 5.352744112903607}} +{"family": "random", "seed": 300007, "x": [0.4971175184934504, 0.4737179460029658, -0.30044941323948327, 0.3550634400349575, 0.4487368841452438, 0.14220531948325865, 0.8421873088538354, 0.6182786364931401, 1.6630744615235187, -1.223861506862476, 0.015479854857265538, 1.1853495692624807, 0.5339634153876449, -0.8532352450747527, 0.9789967773123469, 2.147770522042337, -0.00018578050133811162, 0.15077589927740576, -0.012889017272980474, -0.021584965580598817], "ratio96": 1.982780607293167, "ratio_final": 1.9828824168910333, "diag": {"ratio": 1.9828824168910333, "nonnormal": 1.3143341743805745, "eig_gaps": [0.12229512466217568, 0.2843020159811236, 0.2943428997443752], "discness": 0.02314266615310511, "peaks": 3, "deg": 3, "diamW": 2.7394741967662823}} +{"family": "random", "seed": 300008, "x": [0.8740602120477032, 0.8839503856982609, -0.4922414817391889, -0.35777244525857294, 1.0641312999525254, -0.04895812896610098, -2.416052660890498, -1.4918764488308967, 0.986380727716115, -0.4169070755869627, -2.486460234137772, 1.630953610961164, -0.4494118951814293, 0.3216806216814504, 1.9038235244523956, 1.2988671884278014, -1.9607316571016462, -0.7724398191943074, -0.020266733094105926, -0.020433466240021764], "ratio96": 1.9649935391645037, "ratio_final": 1.9669303506095994, "diag": {"ratio": 1.9669303506095994, "nonnormal": 0.6812694842320294, "eig_gaps": [0.2039461072331789, 0.33989390199414116, 0.39549062881060454], "discness": 0.03382987400832846, "peaks": 3, "deg": 3, "diamW": 4.66826373980073}} +{"family": "random", "seed": 300009, "x": [-0.5363711204223878, -1.2134891963489811, -0.06498936736616445, 1.9229074430760829, -1.3395410644063666, -0.29523171254496783, -0.9785563171384722, -0.6399011644277162, 0.15822049478349975, -1.0156554852927617, 0.9340264581550215, -0.09488945227501336, -2.2730628568454394, 5.5729490795437755, -0.04203608558068267, 0.6141043255679818, -1.4863996268615292, -0.6589378116620452, 3.2082851807202006, 0.04387327515047054], "ratio96": 1.4916871554286917, "ratio_final": 1.4998936239348117, "diag": {"ratio": 1.4998936239348117, "nonnormal": 0.5665072787725145, "eig_gaps": [0.3262590732960073, 0.6841680419867868, 0.848206986769885], "discness": 0.3224291972854775, "peaks": 3, "deg": 3, "diamW": 3.7392073127210765}} +{"family": "random", "seed": 300010, "x": [-0.2599246785450231, 1.3149740839683353, -1.1122434291108116, -0.9652775008580277, 0.7585854684727411, -0.8046856673932159, 2.8303851850281756, 2.4217040619040393, 0.8832971625625394, 1.802435780928307, 0.2236838474592382, 3.204851649062456, -0.6360705075435915, -0.5291770114292194, -0.3603608539326466, 1.1328286103143705, 1.052445780723155, 1.437605805778707, -0.14850086448122732, 0.1246753806901515], "ratio96": 1.9676406517698055, "ratio_final": 1.9676430185679914, "diag": {"ratio": 1.9676430185679914, "nonnormal": 0.7846604045974668, "eig_gaps": [0.3248485370233035, 0.4068439004463389, 0.4211465885996627], "discness": 0.012912215592314613, "peaks": 4, "deg": 3, "diamW": 5.7802595040744364}} +{"family": "random", "seed": 300011, "x": [-1.389248586348946, -0.13289810290739293, 1.1259073580296164, 1.6340402987500575, 0.5422646815419448, 0.019688078001725978, -2.2170539497408566, -5.435430030777681, 1.5650094536448655, 2.5713151623550616, -0.12867636033236013, -0.13503751044322987, 1.8420704154228116, -0.2849771189960923, 1.9082078385384773, 4.620052997280258, -0.0906345250929824, 0.0973893110400646, -0.019189793011511233, 0.0005615459095831504], "ratio96": 1.9663925996580267, "ratio_final": 1.9663806978337794, "diag": {"ratio": 1.9663806978337794, "nonnormal": 1.3586690607004073, "eig_gaps": [0.236209373828989, 0.26660759158420216, 0.4229564661825396], "discness": 0.03870303113435805, "peaks": 3, "deg": 3, "diamW": 7.267350878881991}} +{"family": "random", "seed": 300012, "x": [-0.7926056305264266, -1.1505961102883981, 0.8033814174411891, 0.6697344040139148, -0.19745229162106237, 0.37733143556113546, 1.642121186803954, -1.3190803291797766, -0.23407237980591789, -1.467067121156143, -2.057396802478521, 0.6169618204153873, -0.9584044188894634, -1.3017904302860237, 0.22899673760498174, -0.03442512217194016, 1.857674633273652, -2.0070773222034948, 0.025836432004742088, -0.23413729814538886], "ratio96": 1.9462546977126907, "ratio_final": 1.946212916612244, "diag": {"ratio": 1.946212916612244, "nonnormal": 0.7397553311876851, "eig_gaps": [0.2530887130000983, 0.39801675864627034, 0.5876275578237576], "discness": 0.0753038626186094, "peaks": 3, "deg": 3, "diamW": 4.119793755239455}} diff --git a/problems/crouzeix/data/certified.jsonl b/problems/crouzeix/data/certified.jsonl new file mode 100644 index 0000000..c23e00b --- /dev/null +++ b/problems/crouzeix/data/certified.jsonl @@ -0,0 +1,286 @@ +{"seed": 400000, "family": "random", "class": "stall", "float_ratio": 1.474585189086896, "ratio_lo": "1.435391396029", "ratio_hi_f": 1.8331653518060949, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.81} +{"seed": 401000, "family": "jordan3", "class": "near2-double", "float_ratio": 1.974033911016823, "ratio_lo": "1.915670910227", "ratio_hi_f": 2.01382717113167, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.8} +{"seed": 402000, "family": "jordan2", "class": "near2-other", "float_ratio": 1.972475733235554, "ratio_lo": "1.944390251305", "ratio_hi_f": 2.063970243274458, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 403000, "family": "nearnormal", "class": "stall", "float_ratio": 1.6399400334716803, "ratio_lo": "1.595661049569", "ratio_hi_f": 1.9399626159056633, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.83} +{"seed": 404000, "family": "companion", "class": "stall", "float_ratio": 1.925728641730908, "ratio_lo": "1.893266048203", "ratio_hi_f": 2.025487689886668, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.87} +{"seed": 405000, "family": "peaks", "class": "stall", "float_ratio": 1.9590367342496156, "ratio_lo": "1.874640194300", "ratio_hi_f": 2.2803004806302845, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.85} +{"seed": 406000, "family": "real", "class": "stall", "float_ratio": 1.9563442656411278, "ratio_lo": "1.912316773801", "ratio_hi_f": 2.135791996900198, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.81} +{"seed": 400001, "family": "random", "class": "stall", "float_ratio": 1.9495289236168243, "ratio_lo": "1.900445566313", "ratio_hi_f": 2.1594678692529485, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.84} +{"seed": 401001, "family": "jordan3", "class": "near2-other", "float_ratio": 1.989554662554931, "ratio_lo": "1.912468145160", "ratio_hi_f": 2.1704192426515205, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 402001, "family": "jordan2", "class": "stall", "float_ratio": 1.969392851389826, "ratio_lo": "1.938897699422", "ratio_hi_f": 2.1560824265877154, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.82} +{"seed": 403001, "family": "nearnormal", "class": "stall", "float_ratio": 1.667019965562423, "ratio_lo": "1.645553081557", "ratio_hi_f": 2.0648622925843028, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.82} +{"seed": 404001, "family": "companion", "class": "stall", "float_ratio": 1.9249874758765917, "ratio_lo": "1.867445695800", "ratio_hi_f": 2.3234149736488607, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.88} +{"seed": 405001, "family": "peaks", "class": "near2-other", "float_ratio": 1.987997920438502, "ratio_lo": "1.930768210417", "ratio_hi_f": 2.9982417267958095, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.84} +{"seed": 406001, "family": "real", "class": "near2-double", "float_ratio": 1.987039203566344, "ratio_lo": "1.927647715444", "ratio_hi_f": 2.0677318750285982, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.8} +{"seed": 400002, "family": "random", "class": "stall", "float_ratio": 1.8855779399806656, "ratio_lo": "1.812948812084", "ratio_hi_f": 2.264764237323174, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.88} +{"seed": 401002, "family": "jordan3", "class": "stall", "float_ratio": 1.7552977507389464, "ratio_lo": "1.694044410341", "ratio_hi_f": 1.780573133423289, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.83} +{"seed": 402002, "family": "jordan2", "class": "near2-double", "float_ratio": 1.995382827290676, "ratio_lo": "1.960511448592", "ratio_hi_f": 2.023969803723951, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.8} +{"seed": 403002, "family": "nearnormal", "class": "stall", "float_ratio": 1.8941728389597947, "ratio_lo": "1.859762087136", "ratio_hi_f": 2.0594888396147715, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 404002, "family": "companion", "class": "near2-double", "float_ratio": 1.9756485744848153, "ratio_lo": "1.922058830465", "ratio_hi_f": 2.377811567236033, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.83} +{"seed": 405002, "family": "peaks", "class": "stall", "float_ratio": 1.8892524871320382, "ratio_lo": "1.849144911070", "ratio_hi_f": 2.1139815086641893, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.8} +{"seed": 406002, "family": "real", "class": "stall", "float_ratio": 1.5424697871862874, "ratio_lo": "1.489755481843", "ratio_hi_f": 1.6285274889953685, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.87} +{"seed": 400003, "family": "random", "class": "stall", "float_ratio": 1.3693221045958268, "ratio_lo": "1.348071179552", "ratio_hi_f": 1.525354243574126, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.87} +{"seed": 401003, "family": "jordan3", "class": "stall", "float_ratio": 1.9646022916763135, "ratio_lo": "1.908953198607", "ratio_hi_f": 2.0495296114773827, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.84} +{"seed": 402003, "family": "jordan2", "class": "near2-double", "float_ratio": 1.9942564539334426, "ratio_lo": "1.959657907620", "ratio_hi_f": 2.0484263569273815, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.82} +{"seed": 403003, "family": "nearnormal", "class": "stall", "float_ratio": 1.7126386013132484, "ratio_lo": "1.667414418439", "ratio_hi_f": 1.9292399475296165, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.85} +{"seed": 404003, "family": "companion", "class": "stall", "float_ratio": 1.6764003782028387, "ratio_lo": "1.587832004584", "ratio_hi_f": 1.848466659904112, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.8} +{"seed": 405003, "family": "peaks", "class": "near2-double", "float_ratio": 1.9712314269731561, "ratio_lo": "1.946260999899", "ratio_hi_f": 2.069421276591414, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 406003, "family": "real", "class": "near2-double", "float_ratio": 1.9829369879028436, "ratio_lo": "1.957060281064", "ratio_hi_f": 2.045347336105532, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.81} +{"seed": 400004, "family": "random", "class": "stall", "float_ratio": 1.6791706927654377, "ratio_lo": "1.660958735723", "ratio_hi_f": 2.026261971875639, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.84} +{"seed": 401004, "family": "jordan3", "class": "near2-other", "float_ratio": 1.9912826371971442, "ratio_lo": "1.942675184919", "ratio_hi_f": 2.26122220222037, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.81} +{"seed": 402004, "family": "jordan2", "class": "near2-double", "float_ratio": 1.99108154963197, "ratio_lo": "1.957294081513", "ratio_hi_f": 2.2607085019481317, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.81} +{"seed": 403004, "family": "nearnormal", "class": "one-exact", "float_ratio": 0.9999999999820248, "ratio_lo": "0.999999685121", "ratio_hi_f": 1.0, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.55} +{"seed": 404004, "family": "companion", "class": "stall", "float_ratio": 1.9219565692814136, "ratio_lo": "1.859290456249", "ratio_hi_f": 2.232264476189048, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.88} +{"seed": 405004, "family": "peaks", "class": "near2-double", "float_ratio": 1.9863425155864072, "ratio_lo": "1.941280356866", "ratio_hi_f": 2.0448799001975804, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.84} +{"seed": 406004, "family": "real", "class": "below2-max", "float_ratio": 1.9588423767255672, "ratio_lo": "1.910032463487", "ratio_hi_f": 2.1483211535907287, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.85} +{"seed": 400005, "family": "random", "class": "one-exact", "float_ratio": 0.9999999999752113, "ratio_lo": "0.999999949455", "ratio_hi_f": 1.0, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.54} +{"seed": 401005, "family": "jordan3", "class": "near2-double", "float_ratio": 1.9858703539332918, "ratio_lo": "1.926703933341", "ratio_hi_f": 2.079426855192487, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 402005, "family": "jordan2", "class": "near2-other", "float_ratio": 1.996750219923216, "ratio_lo": "1.962224739100", "ratio_hi_f": 2.0229642806900316, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 403005, "family": "nearnormal", "class": "stall", "float_ratio": 1.4741437723143012, "ratio_lo": "1.428153527812", "ratio_hi_f": 1.7144957361332047, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.8} +{"seed": 404005, "family": "companion", "class": "stall", "float_ratio": 1.6760626219030748, "ratio_lo": "1.615635220377", "ratio_hi_f": 2.20453244282856, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.84} +{"seed": 405005, "family": "peaks", "class": "near2-double", "float_ratio": 1.9828228939265313, "ratio_lo": "1.942032877317", "ratio_hi_f": 2.0366070082542116, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.84} +{"seed": 406005, "family": "real", "class": "stall", "float_ratio": 1.914785202186186, "ratio_lo": "1.857134229824", "ratio_hi_f": 2.0678735844289333, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.81} +{"seed": 400006, "family": "random", "class": "stall", "float_ratio": 1.9603417569805628, "ratio_lo": "1.925854298054", "ratio_hi_f": 2.0995636584247195, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 401006, "family": "jordan3", "class": "near2-other", "float_ratio": 1.9787114916978592, "ratio_lo": "1.938136248368", "ratio_hi_f": 2.0802148161692866, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 402006, "family": "jordan2", "class": "near2-double", "float_ratio": 1.9916104649078281, "ratio_lo": "1.959237861543", "ratio_hi_f": 2.285651582237166, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 403006, "family": "nearnormal", "class": "stall", "float_ratio": 1.9605124786432069, "ratio_lo": "1.847638467808", "ratio_hi_f": 2.0511252630990557, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 404006, "family": "companion", "class": "stall", "float_ratio": 1.8022693302847397, "ratio_lo": "1.774716580576", "ratio_hi_f": 1.9225447274667178, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.81} +{"seed": 405006, "family": "peaks", "class": "stall", "float_ratio": 1.9212837509770775, "ratio_lo": "1.881137577676", "ratio_hi_f": 1.9365150334788037, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 406006, "family": "real", "class": "near2-double", "float_ratio": 1.9763202426022828, "ratio_lo": "1.938127303565", "ratio_hi_f": 2.070208869364135, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 400007, "family": "random", "class": "stall", "float_ratio": 1.4825509958198984, "ratio_lo": "1.442467133113", "ratio_hi_f": 2.067144241555777, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.86} +{"seed": 401007, "family": "jordan3", "class": "near2-double", "float_ratio": 1.9784330459657076, "ratio_lo": "1.925600206226", "ratio_hi_f": 2.0691503132576763, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.82} +{"seed": 402007, "family": "jordan2", "class": "near2-double", "float_ratio": 1.9921654931823096, "ratio_lo": "1.958941826207", "ratio_hi_f": 2.058797868391538, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.7} +{"seed": 403007, "family": "nearnormal", "class": "one-exact", "float_ratio": 0.9999999999855319, "ratio_lo": "0.999999874463", "ratio_hi_f": 1.0, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.53} +{"seed": 404007, "family": "companion", "class": "stall", "float_ratio": 1.9513804021932275, "ratio_lo": "1.894094188128", "ratio_hi_f": 2.2453923651893706, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.83} +{"seed": 405007, "family": "peaks", "class": "stall", "float_ratio": 1.9324690744761903, "ratio_lo": "1.875142190385", "ratio_hi_f": 1.999945403456609, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.83} +{"seed": 406007, "family": "real", "class": "near2-double", "float_ratio": 1.998177627046962, "ratio_lo": "1.964387468289", "ratio_hi_f": 2.02383068589495, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 400008, "family": "random", "class": "near2-other", "float_ratio": 1.971935983053439, "ratio_lo": "1.914614782709", "ratio_hi_f": 2.0642296661676283, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.82} +{"seed": 401008, "family": "jordan3", "class": "near2-other", "float_ratio": 1.9851595972644236, "ratio_lo": "1.929518901820", "ratio_hi_f": 2.231426003270655, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.8} +{"seed": 402008, "family": "jordan2", "class": "near2-double", "float_ratio": 1.9832604521619137, "ratio_lo": "1.956842060444", "ratio_hi_f": 2.097260891807356, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 403008, "family": "nearnormal", "class": "stall", "float_ratio": 1.6624134520100349, "ratio_lo": "1.563801093959", "ratio_hi_f": 1.694067824495038, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 404008, "family": "companion", "class": "stall", "float_ratio": 1.9150984233362012, "ratio_lo": "1.792451493973", "ratio_hi_f": 2.161721706565608, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.88} +{"seed": 405008, "family": "peaks", "class": "stall", "float_ratio": 1.9005921664070224, "ratio_lo": "1.875863301490", "ratio_hi_f": 2.0676406163267855, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.83} +{"seed": 406008, "family": "real", "class": "stall", "float_ratio": 1.7689261204340256, "ratio_lo": "1.731392839397", "ratio_hi_f": 1.8962424355439171, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 400009, "family": "random", "class": "near2-double", "float_ratio": 1.9946369355893576, "ratio_lo": "1.931406420223", "ratio_hi_f": 2.2670434436795635, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 401009, "family": "jordan3", "class": "near2-double", "float_ratio": 1.9947499524268297, "ratio_lo": "1.935292574657", "ratio_hi_f": 2.2632946727799745, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 402009, "family": "jordan2", "class": "near2-other", "float_ratio": 1.9904085835769856, "ratio_lo": "1.962673516764", "ratio_hi_f": 2.0299941703396898, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 403009, "family": "nearnormal", "class": "stall", "float_ratio": 1.7786737363778258, "ratio_lo": "1.711796303477", "ratio_hi_f": 2.072877712638571, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.81} +{"seed": 404009, "family": "companion", "class": "stall", "float_ratio": 1.795200320138665, "ratio_lo": "1.775316069425", "ratio_hi_f": 1.9548201130031182, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 405009, "family": "peaks", "class": "stall", "float_ratio": 1.9657348708525542, "ratio_lo": "1.900785849557", "ratio_hi_f": 2.014356612733733, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 406009, "family": "real", "class": "near2-double", "float_ratio": 1.9986423314570214, "ratio_lo": "1.969724783391", "ratio_hi_f": 2.0514332060563505, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 400010, "family": "random", "class": "stall", "float_ratio": 1.4882500726649115, "ratio_lo": "1.427050814643", "ratio_hi_f": 1.5891523133710337, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.8} +{"seed": 401010, "family": "jordan3", "class": "stall", "float_ratio": 1.9638134664520799, "ratio_lo": "1.903775712933", "ratio_hi_f": 2.110241487832618, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 402010, "family": "jordan2", "class": "stall", "float_ratio": 1.8562662258619247, "ratio_lo": "1.792245892447", "ratio_hi_f": 1.862874561223657, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.65} +{"seed": 403010, "family": "nearnormal", "class": "stall", "float_ratio": 1.8421403904378362, "ratio_lo": "1.770446969281", "ratio_hi_f": 1.9333694115210114, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 404010, "family": "companion", "class": "stall", "float_ratio": 1.9287354498093967, "ratio_lo": "1.863792588464", "ratio_hi_f": 2.0092433529910467, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 405010, "family": "peaks", "class": "near2-other", "float_ratio": 1.9707916175806544, "ratio_lo": "1.939031313648", "ratio_hi_f": 2.0257408033938544, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 406010, "family": "real", "class": "near2-other", "float_ratio": 1.9787659977707426, "ratio_lo": "1.944791285681", "ratio_hi_f": 2.0819536697435614, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 400011, "family": "random", "class": "stall", "float_ratio": 1.9121754407897367, "ratio_lo": "1.781563348003", "ratio_hi_f": 2.6702070672338563, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.86} +{"seed": 401011, "family": "jordan3", "class": "near2-double", "float_ratio": 1.9921639591736189, "ratio_lo": "1.959190758959", "ratio_hi_f": 2.01941718467603, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 500000, "family": "random", "class": "stall", "float_ratio": 1.8845616755755084, "ratio_lo": "1.848739325165", "ratio_hi_f": 2.3707024357846422, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 501000, "family": "jordan3", "class": "near2-double", "float_ratio": 1.979742431611181, "ratio_lo": "1.953861688381", "ratio_hi_f": 2.067912391819757, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 502000, "family": "jordan2", "class": "near2-other", "float_ratio": 1.9725207646608844, "ratio_lo": "1.940903393894", "ratio_hi_f": 2.000574390657588, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 503000, "family": "nearnormal", "class": "one-exact", "float_ratio": 0.9999999999897815, "ratio_lo": "0.999999970196", "ratio_hi_f": 1.0, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.48} +{"seed": 504000, "family": "companion", "class": "stall", "float_ratio": 1.7905543163271889, "ratio_lo": "1.495852522752", "ratio_hi_f": 1.5899709657648975, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.57} +{"seed": 505000, "family": "peaks", "class": "stall", "float_ratio": 1.8785472999010968, "ratio_lo": "1.823568305834", "ratio_hi_f": 2.1515198822997315, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 506000, "family": "real", "class": "stall", "float_ratio": 1.8950095242900347, "ratio_lo": "1.853947019657", "ratio_hi_f": 2.0361876663460867, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 500001, "family": "random", "class": "near2-other", "float_ratio": 1.989647419435718, "ratio_lo": "1.960162145884", "ratio_hi_f": 2.015087173911618, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 501001, "family": "jordan3", "class": "near2-double", "float_ratio": 1.9953155484498082, "ratio_lo": "1.939269448303", "ratio_hi_f": 2.1017196845601447, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 502001, "family": "jordan2", "class": "near2-other", "float_ratio": 1.9835924689440168, "ratio_lo": "1.955673234514", "ratio_hi_f": 2.0612196624507204, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 503001, "family": "nearnormal", "class": "stall", "float_ratio": 1.2480859785227605, "ratio_lo": "1.230906478043", "ratio_hi_f": 1.3441746886776256, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.67} +{"seed": 504001, "family": "companion", "class": "stall", "float_ratio": 1.9198552462314853, "ratio_lo": "1.838914028549", "ratio_hi_f": 2.329218977672877, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.8} +{"seed": 505001, "family": "peaks", "class": "near2-other", "float_ratio": 1.9919948550362034, "ratio_lo": "1.934722614602", "ratio_hi_f": 2.2482459703385325, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 506001, "family": "real", "class": "near2-double", "float_ratio": 1.9986293893348943, "ratio_lo": "1.957962481957", "ratio_hi_f": 2.04741846022307, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.6} +{"seed": 500002, "family": "random", "class": "near2-double", "float_ratio": 1.9825954160573804, "ratio_lo": "1.932974352842", "ratio_hi_f": 2.2470360354907406, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 501002, "family": "jordan3", "class": "near2-other", "float_ratio": 1.9746841950990968, "ratio_lo": "1.917128182332", "ratio_hi_f": 2.043333701649846, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 502002, "family": "jordan2", "class": "stall", "float_ratio": 1.8437007000903027, "ratio_lo": "1.820083874166", "ratio_hi_f": 1.8451746214117075, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 503002, "family": "nearnormal", "class": "stall", "float_ratio": 1.5768934243014407, "ratio_lo": "1.560839149182", "ratio_hi_f": 1.6099049142423996, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 504002, "family": "companion", "class": "one-exact", "float_ratio": 0.9999999999848634, "ratio_lo": "0.999999884019", "ratio_hi_f": 1.0, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.52} +{"seed": 505002, "family": "peaks", "class": "near2-other", "float_ratio": 1.981347870661632, "ratio_lo": "1.932247648046", "ratio_hi_f": 2.0680820922813017, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 506002, "family": "real", "class": "stall", "float_ratio": 1.287282698565298, "ratio_lo": "1.259210507607", "ratio_hi_f": 1.505682623462959, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 500003, "family": "random", "class": "stall", "float_ratio": 1.9359821267482429, "ratio_lo": "1.889461979083", "ratio_hi_f": 2.1223083591854266, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 501003, "family": "jordan3", "class": "stall", "float_ratio": 1.9594882050402616, "ratio_lo": "1.905079793818", "ratio_hi_f": 2.3182354513672556, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 502003, "family": "jordan2", "class": "near2-other", "float_ratio": 1.972273657766895, "ratio_lo": "1.945065940978", "ratio_hi_f": 2.005313743406287, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 503003, "family": "nearnormal", "class": "stall", "float_ratio": 1.8592570938725468, "ratio_lo": "1.838640594431", "ratio_hi_f": 1.9453400429945327, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 504003, "family": "companion", "class": "stall", "float_ratio": 1.5720460083945795, "ratio_lo": "1.542323659765", "ratio_hi_f": 1.6453034563313724, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 505003, "family": "peaks", "class": "stall", "float_ratio": 1.8616595183047597, "ratio_lo": "1.837091054935", "ratio_hi_f": 1.8981148585508145, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 506003, "family": "real", "class": "stall", "float_ratio": 1.8660990242562745, "ratio_lo": "1.801020011987", "ratio_hi_f": 2.0498817378774055, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 500004, "family": "random", "class": "stall", "float_ratio": 1.7182643773409472, "ratio_lo": "1.617571254769", "ratio_hi_f": 1.9223608771549197, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 501004, "family": "jordan3", "class": "near2-other", "float_ratio": 1.971277939419618, "ratio_lo": "1.906905698135", "ratio_hi_f": 2.0640591070189913, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 502004, "family": "jordan2", "class": "near2-double", "float_ratio": 1.972615867427874, "ratio_lo": "1.942768247245", "ratio_hi_f": 2.0516934133061313, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 503004, "family": "nearnormal", "class": "stall", "float_ratio": 1.7991307931216087, "ratio_lo": "1.372140147297", "ratio_hi_f": 1.504405119302281, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.55} +{"seed": 504004, "family": "companion", "class": "stall", "float_ratio": 1.8727094280973273, "ratio_lo": "1.855763078301", "ratio_hi_f": 2.3127699678751372, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 505004, "family": "peaks", "class": "stall", "float_ratio": 1.9439309034314967, "ratio_lo": "1.911578489051", "ratio_hi_f": 2.031980403976008, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 506004, "family": "real", "class": "stall", "float_ratio": 1.8437241614229833, "ratio_lo": "1.804497867040", "ratio_hi_f": 1.9435438841802255, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 500005, "family": "random", "class": "stall", "float_ratio": 1.0229491721467197, "ratio_lo": "1.022559544568", "ratio_hi_f": 1.067863293458394, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 501005, "family": "jordan3", "class": "near2-other", "float_ratio": 1.991434563410379, "ratio_lo": "1.914821676728", "ratio_hi_f": 2.034338722470689, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 502005, "family": "jordan2", "class": "near2-double", "float_ratio": 1.9900582366896558, "ratio_lo": "1.962222124095", "ratio_hi_f": 2.152162227902933, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 503005, "family": "nearnormal", "class": "stall", "float_ratio": 1.5129410425167429, "ratio_lo": "1.502120092208", "ratio_hi_f": 1.741717883273398, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 504005, "family": "companion", "class": "stall", "float_ratio": 1.6276224592666562, "ratio_lo": "1.569591053622", "ratio_hi_f": 2.576422904651885, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 505005, "family": "peaks", "class": "stall", "float_ratio": 1.9123656411555223, "ratio_lo": "1.870064207329", "ratio_hi_f": 2.1130298855008482, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 506005, "family": "real", "class": "stall", "float_ratio": 1.913455138646116, "ratio_lo": "1.886296198641", "ratio_hi_f": 2.0264500985413925, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 500006, "family": "random", "class": "stall", "float_ratio": 1.8827341824610222, "ratio_lo": "1.836990030986", "ratio_hi_f": 2.017179150475628, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 501006, "family": "jordan3", "class": "stall", "float_ratio": 1.9559317390935447, "ratio_lo": "1.922114957644", "ratio_hi_f": 2.0842409812272678, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 502006, "family": "jordan2", "class": "near2-double", "float_ratio": 1.9878657349927435, "ratio_lo": "1.958452603457", "ratio_hi_f": 2.117007365069295, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 503006, "family": "nearnormal", "class": "stall", "float_ratio": 1.603390187982936, "ratio_lo": "1.518845023772", "ratio_hi_f": 1.6305780382986232, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.7} +{"seed": 504006, "family": "companion", "class": "stall", "float_ratio": 1.8219698678474072, "ratio_lo": "1.769006307135", "ratio_hi_f": 1.957628866489152, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 505006, "family": "peaks", "class": "stall", "float_ratio": 1.9423007722174221, "ratio_lo": "1.911127726080", "ratio_hi_f": 2.0543127280744717, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 506006, "family": "real", "class": "stall", "float_ratio": 1.107756656457773, "ratio_lo": "1.096113265937", "ratio_hi_f": 1.1506037834913039, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 500007, "family": "random", "class": "stall", "float_ratio": 1.6783193414620063, "ratio_lo": "1.626248827787", "ratio_hi_f": 1.968090600602731, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 501007, "family": "jordan3", "class": "stall", "float_ratio": 1.9609974357998128, "ratio_lo": "1.918457352702", "ratio_hi_f": 2.142550499098634, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 502007, "family": "jordan2", "class": "near2-double", "float_ratio": 1.98673415332121, "ratio_lo": "1.951295938974", "ratio_hi_f": 2.1939809604219307, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.69} +{"seed": 503007, "family": "nearnormal", "class": "stall", "float_ratio": 1.311502965762841, "ratio_lo": "1.303244615521", "ratio_hi_f": 1.587763266979006, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 504007, "family": "companion", "class": "stall", "float_ratio": 1.8730875103984932, "ratio_lo": "1.798733828260", "ratio_hi_f": 2.1557663079343645, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 505007, "family": "peaks", "class": "stall", "float_ratio": 1.8600022210979095, "ratio_lo": "1.824629580814", "ratio_hi_f": 1.8971032880106902, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 506007, "family": "real", "class": "near2-double", "float_ratio": 1.9896822883667182, "ratio_lo": "1.951145198193", "ratio_hi_f": 2.077928043936996, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 500008, "family": "random", "class": "stall", "float_ratio": 1.9684583081604405, "ratio_lo": "1.906507167891", "ratio_hi_f": 2.169697875832118, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 501008, "family": "jordan3", "class": "near2-other", "float_ratio": 1.9779558369327876, "ratio_lo": "1.928746175877", "ratio_hi_f": 2.018332730872004, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 502008, "family": "jordan2", "class": "stall", "float_ratio": 1.9540779197750053, "ratio_lo": "1.922011026319", "ratio_hi_f": 2.08595433183867, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 503008, "family": "nearnormal", "class": "stall", "float_ratio": 1.9353272570811622, "ratio_lo": "1.902393029045", "ratio_hi_f": 2.0276294359767126, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 504008, "family": "companion", "class": "stall", "float_ratio": 1.9146148700527859, "ratio_lo": "1.846468129481", "ratio_hi_f": 2.044103090462988, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.67} +{"seed": 505008, "family": "peaks", "class": "near2-other", "float_ratio": 1.9802751031818326, "ratio_lo": "1.943255652452", "ratio_hi_f": 2.0335708068892835, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 506008, "family": "real", "class": "near2-other", "float_ratio": 1.973556833656473, "ratio_lo": "1.904042656279", "ratio_hi_f": 1.9930727802317116, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 500009, "family": "random", "class": "one-exact", "float_ratio": 0.9999999999999187, "ratio_lo": "0.999999999882", "ratio_hi_f": 1.0, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.5} +{"seed": 501009, "family": "jordan3", "class": "near2-double", "float_ratio": 1.9852625351617286, "ratio_lo": "1.932557635576", "ratio_hi_f": 2.097594942268014, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 502009, "family": "jordan2", "class": "near2-double", "float_ratio": 1.9773480551859004, "ratio_lo": "1.948200482592", "ratio_hi_f": 2.027176593449858, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.7} +{"seed": 503009, "family": "nearnormal", "class": "stall", "float_ratio": 1.176328823444026, "ratio_lo": "1.165214033208", "ratio_hi_f": 1.2742760059282043, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.7} +{"seed": 504009, "family": "companion", "class": "stall", "float_ratio": 1.9352691803344193, "ratio_lo": "1.893809117731", "ratio_hi_f": 2.069835708586335, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 505009, "family": "peaks", "class": "stall", "float_ratio": 1.9697441665215818, "ratio_lo": "1.934094464917", "ratio_hi_f": 2.189321030359636, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 506009, "family": "real", "class": "stall", "float_ratio": 1.7520955385073727, "ratio_lo": "1.547554040835", "ratio_hi_f": 1.6673139317737382, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.6} +{"seed": 500010, "family": "random", "class": "stall", "float_ratio": 1.45736747551724, "ratio_lo": "1.411907752445", "ratio_hi_f": 1.6796245083164862, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 501010, "family": "jordan3", "class": "near2-other", "float_ratio": 1.9915603286186754, "ratio_lo": "1.943589779339", "ratio_hi_f": 2.1296729065122464, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 502010, "family": "jordan2", "class": "near2-double", "float_ratio": 1.9726171510423114, "ratio_lo": "1.943834910457", "ratio_hi_f": 2.02004992978582, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 503010, "family": "nearnormal", "class": "one-exact", "float_ratio": 0.9999999999783298, "ratio_lo": "0.999999797046", "ratio_hi_f": 1.0, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.47} +{"seed": 504010, "family": "companion", "class": "below2-max", "float_ratio": 1.7811672748407623, "ratio_lo": "1.754874681196", "ratio_hi_f": 1.8483103413998854, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 505010, "family": "peaks", "class": "stall", "float_ratio": 1.9390347714661518, "ratio_lo": "1.863647091795", "ratio_hi_f": 2.2163205966673583, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 506010, "family": "real", "class": "stall", "float_ratio": 1.92459646692251, "ratio_lo": "1.870857956782", "ratio_hi_f": 2.1372560729903944, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.67} +{"seed": 500011, "family": "random", "class": "stall", "float_ratio": 1.9468717506582869, "ratio_lo": "1.901401194959", "ratio_hi_f": 1.9679889674769813, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 501011, "family": "jordan3", "class": "stall", "float_ratio": 1.8942031515403743, "ratio_lo": "1.847359077244", "ratio_hi_f": 2.317485957259326, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 502011, "family": "jordan2", "class": "near2-other", "float_ratio": 1.9919396766211583, "ratio_lo": "1.954709577283", "ratio_hi_f": 2.0493360695731972, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 503011, "family": "nearnormal", "class": "one-exact", "float_ratio": 0.9999999999789269, "ratio_lo": "0.999999830285", "ratio_hi_f": 1.0, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.48} +{"seed": 504011, "family": "companion", "class": "stall", "float_ratio": 1.5434588388660222, "ratio_lo": "1.415836610503", "ratio_hi_f": 1.6871869845013403, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 505011, "family": "peaks", "class": "stall", "float_ratio": 1.9608644336249832, "ratio_lo": "1.916338999894", "ratio_hi_f": 2.0999563558958725, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 600000, "family": "random", "class": "near2-other", "float_ratio": 1.981859301889913, "ratio_lo": "1.921192291820", "ratio_hi_f": 2.239860562193161, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.82} +{"seed": 601000, "family": "jordan3", "class": "near2-double", "float_ratio": 1.989449813565899, "ratio_lo": "1.927003533416", "ratio_hi_f": 2.172701681837788, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 602000, "family": "jordan2", "class": "near2-other", "float_ratio": 1.9825064773241765, "ratio_lo": "1.955314567171", "ratio_hi_f": 2.033674361977302, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 603000, "family": "nearnormal", "class": "stall", "float_ratio": 1.7996378111063278, "ratio_lo": "1.747105277867", "ratio_hi_f": 1.9142575304792697, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 604000, "family": "companion", "class": "stall", "float_ratio": 1.8609734670980096, "ratio_lo": "1.837027446747", "ratio_hi_f": 1.9913552400102568, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 605000, "family": "peaks", "class": "stall", "float_ratio": 1.8222197897543364, "ratio_lo": "1.474158815591", "ratio_hi_f": 1.877716061315971, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 606000, "family": "real", "class": "stall", "float_ratio": 1.8214054701696325, "ratio_lo": "1.740209612833", "ratio_hi_f": 2.054839927227051, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 600001, "family": "random", "class": "stall", "float_ratio": 1.854996819679032, "ratio_lo": "1.739657986460", "ratio_hi_f": 2.0605933586891467, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 601001, "family": "jordan3", "class": "near2-other", "float_ratio": 1.9746157296609834, "ratio_lo": "1.917171223685", "ratio_hi_f": 1.9825219071819864, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 602001, "family": "jordan2", "class": "near2-other", "float_ratio": 1.9845199763927919, "ratio_lo": "1.953215631020", "ratio_hi_f": 2.029240761577828, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 603001, "family": "nearnormal", "class": "stall", "float_ratio": 1.7640525690219728, "ratio_lo": "1.618886734415", "ratio_hi_f": 1.8120956413222344, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 604001, "family": "companion", "class": "stall", "float_ratio": 1.3853016218976717, "ratio_lo": "1.343266780515", "ratio_hi_f": 1.4825991964002827, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 605001, "family": "peaks", "class": "near2-other", "float_ratio": 1.9704061233923598, "ratio_lo": "1.935374805874", "ratio_hi_f": 1.9827539304711732, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 606001, "family": "real", "class": "near2-double", "float_ratio": 1.990434852368694, "ratio_lo": "1.954726418405", "ratio_hi_f": 2.3712623908469297, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 600002, "family": "random", "class": "stall", "float_ratio": 1.5989092253951127, "ratio_lo": "1.512474174367", "ratio_hi_f": 2.2994610735882786, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 601002, "family": "jordan3", "class": "near2-other", "float_ratio": 1.970341072659899, "ratio_lo": "1.923230650837", "ratio_hi_f": 2.241034161155233, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 602002, "family": "jordan2", "class": "near2-other", "float_ratio": 1.9930091694590535, "ratio_lo": "1.961554546100", "ratio_hi_f": 2.0040335608842983, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 603002, "family": "nearnormal", "class": "near2-double", "float_ratio": 1.9718211928815372, "ratio_lo": "1.928505749023", "ratio_hi_f": 1.983872953615661, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.67} +{"seed": 604002, "family": "companion", "class": "below2-max", "float_ratio": 1.96025972749273, "ratio_lo": "1.894167519159", "ratio_hi_f": 2.114330788036722, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 605002, "family": "peaks", "class": "near2-other", "float_ratio": 1.9702663383359411, "ratio_lo": "1.896857768378", "ratio_hi_f": 2.2697861596612126, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 606002, "family": "real", "class": "near2-double", "float_ratio": 1.9879277920417806, "ratio_lo": "1.956386024767", "ratio_hi_f": 2.0368682415993136, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 600003, "family": "random", "class": "stall", "float_ratio": 1.8970449526308026, "ratio_lo": "1.835413711014", "ratio_hi_f": 2.1472420510144397, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 601003, "family": "jordan3", "class": "near2-other", "float_ratio": 1.994816612552754, "ratio_lo": "1.942366731602", "ratio_hi_f": 2.229217371276228, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 602003, "family": "jordan2", "class": "stall", "float_ratio": 1.9316633108665184, "ratio_lo": "1.899952698534", "ratio_hi_f": 2.160387056935884, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 603003, "family": "nearnormal", "class": "stall", "float_ratio": 1.7187594873165428, "ratio_lo": "1.321095828838", "ratio_hi_f": 1.5338750236738645, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.64} +{"seed": 604003, "family": "companion", "class": "stall", "float_ratio": 1.9462486629760212, "ratio_lo": "1.885830398951", "ratio_hi_f": 2.0756353848914766, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.67} +{"seed": 605003, "family": "peaks", "class": "stall", "float_ratio": 1.9504934396249074, "ratio_lo": "1.899423573859", "ratio_hi_f": 2.112677385040754, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 606003, "family": "real", "class": "stall", "float_ratio": 1.9289403584145925, "ratio_lo": "1.901050901634", "ratio_hi_f": 2.0854943222228672, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.68} +{"seed": 600004, "family": "random", "class": "stall", "float_ratio": 1.9349931508332516, "ratio_lo": "1.886145165126", "ratio_hi_f": 2.294920263031434, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 601004, "family": "jordan3", "class": "near2-triple", "float_ratio": 1.995203391499658, "ratio_lo": "1.934241437245", "ratio_hi_f": 2.1184892451602924, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 602004, "family": "jordan2", "class": "near2-double", "float_ratio": 1.9966914813962127, "ratio_lo": "1.965527148518", "ratio_hi_f": 2.0479545029772486, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 603004, "family": "nearnormal", "class": "one-exact", "float_ratio": 0.9999999999855295, "ratio_lo": "0.992963116512", "ratio_hi_f": 1.0, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.53} +{"seed": 604004, "family": "companion", "class": "stall", "float_ratio": 1.5758094914744551, "ratio_lo": "1.549312175272", "ratio_hi_f": 1.8406163281778967, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 605004, "family": "peaks", "class": "near2-other", "float_ratio": 1.978559453882369, "ratio_lo": "1.926622039714", "ratio_hi_f": 2.094948180872082, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 606004, "family": "real", "class": "stall", "float_ratio": 1.5463886315837072, "ratio_lo": "1.521299037575", "ratio_hi_f": 1.6922536674015989, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 600005, "family": "random", "class": "stall", "float_ratio": 1.93390661994052, "ratio_lo": "1.888570272083", "ratio_hi_f": 2.460314104884876, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 601005, "family": "jordan3", "class": "near2-other", "float_ratio": 1.9799329149467628, "ratio_lo": "1.917057006117", "ratio_hi_f": 2.526453640371936, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 602005, "family": "jordan2", "class": "near2-other", "float_ratio": 1.9848139471528905, "ratio_lo": "1.955027445017", "ratio_hi_f": 2.0333560743363446, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 603005, "family": "nearnormal", "class": "stall", "float_ratio": 1.961538994240155, "ratio_lo": "1.926253285599", "ratio_hi_f": 2.107823154114415, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 604005, "family": "companion", "class": "stall", "float_ratio": 1.4635977531638367, "ratio_lo": "1.453674410716", "ratio_hi_f": 1.493145786287809, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 605005, "family": "peaks", "class": "stall", "float_ratio": 1.8420652209399841, "ratio_lo": "1.760146516958", "ratio_hi_f": 1.9182642485895174, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 606005, "family": "real", "class": "stall", "float_ratio": 1.944038870107737, "ratio_lo": "1.913118156627", "ratio_hi_f": 1.964442392361428, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.8} +{"seed": 600006, "family": "random", "class": "stall", "float_ratio": 1.9153715426347542, "ratio_lo": "1.884035888275", "ratio_hi_f": 2.1612243540441076, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 601006, "family": "jordan3", "class": "near2-other", "float_ratio": 1.9967694422831415, "ratio_lo": "1.925986825953", "ratio_hi_f": 2.072421238448219, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 602006, "family": "jordan2", "class": "near2-double", "float_ratio": 1.993105993398295, "ratio_lo": "1.962471629481", "ratio_hi_f": 2.335548280784641, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 603006, "family": "nearnormal", "class": "stall", "float_ratio": 1.605585731404586, "ratio_lo": "1.452284405529", "ratio_hi_f": 1.6507330113693905, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 604006, "family": "companion", "class": "stall", "float_ratio": 1.6609233832027015, "ratio_lo": "1.489071285640", "ratio_hi_f": 1.5623233576242435, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.7} +{"seed": 605006, "family": "peaks", "class": "stall", "float_ratio": 1.9532009263603356, "ratio_lo": "1.897518792487", "ratio_hi_f": 2.4942347757768175, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 606006, "family": "real", "class": "stall", "float_ratio": 1.6270231550005925, "ratio_lo": "1.549770382092", "ratio_hi_f": 1.7253845399998538, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 600007, "family": "random", "class": "stall", "float_ratio": 1.8869796101928276, "ratio_lo": "1.835103043739", "ratio_hi_f": 2.2372113941895333, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 601007, "family": "jordan3", "class": "stall", "float_ratio": 1.8974921440997967, "ratio_lo": "1.835670537258", "ratio_hi_f": 2.0735399732568327, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 602007, "family": "jordan2", "class": "near2-triple", "float_ratio": 1.998750459681895, "ratio_lo": "1.963852765627", "ratio_hi_f": 2.0031389201004695, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 603007, "family": "nearnormal", "class": "one-exact", "float_ratio": 0.9999999999797999, "ratio_lo": "0.999999769329", "ratio_hi_f": 1.0, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.48} +{"seed": 604007, "family": "companion", "class": "stall", "float_ratio": 1.6494641836904647, "ratio_lo": "1.551220697521", "ratio_hi_f": 1.778645196374973, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 605007, "family": "peaks", "class": "near2-double", "float_ratio": 1.9983602386976598, "ratio_lo": "1.926184303129", "ratio_hi_f": 2.3989497020460173, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.82} +{"seed": 606007, "family": "real", "class": "near2-double", "float_ratio": 1.9963184216629908, "ratio_lo": "1.965215003814", "ratio_hi_f": 2.0970240098074187, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 600008, "family": "random", "class": "stall", "float_ratio": 1.8890855571076424, "ratio_lo": "1.850003480994", "ratio_hi_f": 1.893634864948454, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 601008, "family": "jordan3", "class": "near2-double", "float_ratio": 1.9881820015402822, "ratio_lo": "1.938389794654", "ratio_hi_f": 2.2197576787024067, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 602008, "family": "jordan2", "class": "stall", "float_ratio": 1.9372096446727198, "ratio_lo": "1.908672358257", "ratio_hi_f": 2.006565590817862, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.64} +{"seed": 603008, "family": "nearnormal", "class": "stall", "float_ratio": 1.9670049320670957, "ratio_lo": "1.924657315842", "ratio_hi_f": 2.0614394357629817, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 604008, "family": "companion", "class": "stall", "float_ratio": 1.622248131205978, "ratio_lo": "1.589050963019", "ratio_hi_f": 1.6595919027418318, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 605008, "family": "peaks", "class": "stall", "float_ratio": 1.744658482946865, "ratio_lo": "1.628556165986", "ratio_hi_f": 1.8414075776582903, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.68} +{"seed": 606008, "family": "real", "class": "one-exact", "float_ratio": 0.9999999999771068, "ratio_lo": "0.999999998136", "ratio_hi_f": 1.0, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.45} +{"seed": 600009, "family": "random", "class": "stall", "float_ratio": 1.9457992572343816, "ratio_lo": "1.919180287502", "ratio_hi_f": 2.006987675979092, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 601009, "family": "jordan3", "class": "near2-double", "float_ratio": 1.9930006270504441, "ratio_lo": "1.929298404329", "ratio_hi_f": 2.0676646454014236, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 602009, "family": "jordan2", "class": "stall", "float_ratio": 1.9692063347728552, "ratio_lo": "1.933326908030", "ratio_hi_f": 2.067668647635181, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 603009, "family": "nearnormal", "class": "one-exact", "float_ratio": 0.9999999999813017, "ratio_lo": "0.999999778032", "ratio_hi_f": 1.0, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.49} +{"seed": 604009, "family": "companion", "class": "stall", "float_ratio": 1.6488330511479956, "ratio_lo": "1.608056779631", "ratio_hi_f": 1.852583125355049, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 605009, "family": "peaks", "class": "stall", "float_ratio": 1.923676024962603, "ratio_lo": "1.883843355376", "ratio_hi_f": 1.9614733154178563, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 606009, "family": "real", "class": "stall", "float_ratio": 1.6579877364104212, "ratio_lo": "1.498760350248", "ratio_hi_f": 1.6203755442173091, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 600010, "family": "random", "class": "stall", "float_ratio": 1.943184189433361, "ratio_lo": "1.906399668021", "ratio_hi_f": 2.4845073359937535, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 601010, "family": "jordan3", "class": "near2-other", "float_ratio": 1.977700890059782, "ratio_lo": "1.913440620603", "ratio_hi_f": 2.156624876270276, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 602010, "family": "jordan2", "class": "near2-other", "float_ratio": 1.9829410234617277, "ratio_lo": "1.946757982377", "ratio_hi_f": 2.0591208921923805, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 603010, "family": "nearnormal", "class": "stall", "float_ratio": 1.8307680805342925, "ratio_lo": "1.778472433996", "ratio_hi_f": 1.9129970532262313, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 604010, "family": "companion", "class": "stall", "float_ratio": 1.0027249091450632, "ratio_lo": "1.002077053265", "ratio_hi_f": 1.0226696320838033, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 605010, "family": "peaks", "class": "stall", "float_ratio": 1.954813825118366, "ratio_lo": "1.913225107144", "ratio_hi_f": 2.118407038785451, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 606010, "family": "real", "class": "near2-other", "float_ratio": 1.9801538184736822, "ratio_lo": "1.956067956175", "ratio_hi_f": 2.0075190981033377, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 600011, "family": "random", "class": "stall", "float_ratio": 1.91721205478952, "ratio_lo": "1.868574295679", "ratio_hi_f": 2.185981873527602, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 601011, "family": "jordan3", "class": "stall", "float_ratio": 1.9694424988248191, "ratio_lo": "1.915636082918", "ratio_hi_f": 2.0348793119504958, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.65} +{"seed": 100000, "family": "random", "class": "near2-other", "float_ratio": 1.9730790509888327, "ratio_lo": "1.913504888341", "ratio_hi_f": 3.028132787222701, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 100001, "family": "random", "class": "stall", "float_ratio": 1.9454260507662304, "ratio_lo": "1.897992555973", "ratio_hi_f": 2.2041974787478065, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 100002, "family": "random", "class": "stall", "float_ratio": 1.9658338189373585, "ratio_lo": "1.905593722716", "ratio_hi_f": 2.115697158206725, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 100003, "family": "random", "class": "one-exact", "float_ratio": 0.9999999999999278, "ratio_lo": "0.999999962722", "ratio_hi_f": 1.0, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.51} +{"seed": 100004, "family": "random", "class": "stall", "float_ratio": 1.8343911617633484, "ratio_lo": "1.804437095568", "ratio_hi_f": 1.9481808385399577, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 100005, "family": "random", "class": "stall", "float_ratio": 1.0613982600645762, "ratio_lo": "1.059392485312", "ratio_hi_f": 1.066894444477114, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 100006, "family": "random", "class": "stall", "float_ratio": 1.6810540159458285, "ratio_lo": "1.375376527068", "ratio_hi_f": 1.7613245426039537, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.69} +{"seed": 100007, "family": "random", "class": "stall", "float_ratio": 1.5277950666639002, "ratio_lo": "1.405298630766", "ratio_hi_f": 1.5589566776814214, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.72} +{"seed": 100008, "family": "random", "class": "near2-other", "float_ratio": 1.973967535275769, "ratio_lo": "1.940668511639", "ratio_hi_f": 1.9992254125437334, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 100009, "family": "random", "class": "stall", "float_ratio": 1.942288718514082, "ratio_lo": "1.893788783887", "ratio_hi_f": 2.6338788900060877, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 100010, "family": "random", "class": "stall", "float_ratio": 1.4999138730193975, "ratio_lo": "1.486047958319", "ratio_hi_f": 1.728552303406965, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 100011, "family": "random", "class": "stall", "float_ratio": 1.8080969842886498, "ratio_lo": "1.755559711268", "ratio_hi_f": 2.7027936714321754, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.81} +{"seed": 100012, "family": "random", "class": "stall", "float_ratio": 1.8764272748780273, "ratio_lo": "1.792415219481", "ratio_hi_f": 1.8663273439925476, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 100013, "family": "random", "class": "stall", "float_ratio": 1.6031759311287956, "ratio_lo": "1.584812688385", "ratio_hi_f": 1.7633335118936606, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 100014, "family": "random", "class": "near2-double", "float_ratio": 1.9885905206198318, "ratio_lo": "1.939226686756", "ratio_hi_f": 2.547911347192544, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 100015, "family": "random", "class": "stall", "float_ratio": 1.3024986626469275, "ratio_lo": "1.268921721086", "ratio_hi_f": 1.543183695714946, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.97} +{"seed": 100016, "family": "random", "class": "stall", "float_ratio": 1.5670927509184347, "ratio_lo": "1.544258842312", "ratio_hi_f": 2.0367278396485213, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 100017, "family": "random", "class": "stall", "float_ratio": 1.6198671741374246, "ratio_lo": "1.589527696544", "ratio_hi_f": 1.799062755969996, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.73} +{"seed": 100018, "family": "random", "class": "stall", "float_ratio": 1.4605379236497829, "ratio_lo": "1.425213918823", "ratio_hi_f": 1.6972275910956642, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 200000, "family": "random", "class": "stall", "float_ratio": 1.8910074964009838, "ratio_lo": "1.860921763806", "ratio_hi_f": 2.0511153179313135, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 200001, "family": "random", "class": "stall", "float_ratio": 1.6447805433383824, "ratio_lo": "1.609776491630", "ratio_hi_f": 1.7208589327965729, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 200002, "family": "random", "class": "stall", "float_ratio": 1.7372606665458985, "ratio_lo": "1.667213350305", "ratio_hi_f": 1.8617642895913902, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 200003, "family": "random", "class": "near2-double", "float_ratio": 1.9826930963618628, "ratio_lo": "1.903456562772", "ratio_hi_f": 2.2939481025769797, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 200004, "family": "random", "class": "stall", "float_ratio": 1.688628520465699, "ratio_lo": "1.639048918558", "ratio_hi_f": 1.9224029422766704, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 200005, "family": "random", "class": "stall", "float_ratio": 1.7400715415585755, "ratio_lo": "1.693799732289", "ratio_hi_f": 1.9798196707577012, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 200006, "family": "random", "class": "stall", "float_ratio": 1.9320959989558406, "ratio_lo": "1.901188348735", "ratio_hi_f": 1.9787930027152467, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.74} +{"seed": 200007, "family": "random", "class": "stall", "float_ratio": 1.8404060520565506, "ratio_lo": "1.803119930036", "ratio_hi_f": 2.074349865235065, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 200008, "family": "random", "class": "stall", "float_ratio": 1.9606714033995576, "ratio_lo": "1.910801589628", "ratio_hi_f": 2.5798138575740848, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 200009, "family": "random", "class": "stall", "float_ratio": 1.9624579595306595, "ratio_lo": "1.910962209659", "ratio_hi_f": 2.395013060659718, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 200010, "family": "random", "class": "near2-other", "float_ratio": 1.993412142555695, "ratio_lo": "1.927725483236", "ratio_hi_f": 2.7601733071567165, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.83} +{"seed": 200011, "family": "random", "class": "stall", "float_ratio": 1.9347119896752438, "ratio_lo": "1.877269989287", "ratio_hi_f": 2.3977800585989537, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 200012, "family": "random", "class": "near2-double", "float_ratio": 1.9772965831464346, "ratio_lo": "1.907824108640", "ratio_hi_f": 1.9804274003358473, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.8} +{"seed": 300000, "family": "random", "class": "stall", "float_ratio": 1.4918182509573383, "ratio_lo": "1.444297891546", "ratio_hi_f": 1.689634199396984, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 300001, "family": "random", "class": "stall", "float_ratio": 1.6745373486800608, "ratio_lo": "1.640832804086", "ratio_hi_f": 1.7252551499024777, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 300002, "family": "random", "class": "stall", "float_ratio": 1.7273642851384579, "ratio_lo": "1.676040115795", "ratio_hi_f": 1.83249816249798, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 300003, "family": "random", "class": "near2-double", "float_ratio": 1.988703121753976, "ratio_lo": "1.954375038655", "ratio_hi_f": 2.0048680423580207, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.71} +{"seed": 300004, "family": "random", "class": "stall", "float_ratio": 1.650451247896331, "ratio_lo": "1.603115131783", "ratio_hi_f": 1.728563402606663, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.75} +{"seed": 300005, "family": "random", "class": "stall", "float_ratio": 1.952715171666236, "ratio_lo": "1.906159818784", "ratio_hi_f": 2.0836322975884043, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 300006, "family": "random", "class": "stall", "float_ratio": 1.9236629681671338, "ratio_lo": "1.890115882637", "ratio_hi_f": 2.2409167525222915, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 300007, "family": "random", "class": "near2-double", "float_ratio": 1.9828824168910333, "ratio_lo": "1.952164634537", "ratio_hi_f": 2.007207433165584, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} +{"seed": 300008, "family": "random", "class": "stall", "float_ratio": 1.9669303506095994, "ratio_lo": "1.920034421168", "ratio_hi_f": 2.681342550194931, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.78} +{"seed": 300009, "family": "random", "class": "stall", "float_ratio": 1.4998936239348117, "ratio_lo": "1.466057606078", "ratio_hi_f": 1.6517378657084747, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.76} +{"seed": 300010, "family": "random", "class": "below2-max", "float_ratio": 1.9676430185679914, "ratio_lo": "1.919347816113", "ratio_hi_f": 2.1797946694067907, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.83} +{"seed": 300011, "family": "random", "class": "stall", "float_ratio": 1.9663806978337794, "ratio_lo": "1.926494411466", "ratio_hi_f": 2.0007380628978573, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.79} +{"seed": 300012, "family": "random", "class": "stall", "float_ratio": 1.946212916612244, "ratio_lo": "1.898056340188", "ratio_hi_f": 2.03603682843731, "refutes": false, "den": 256, "dirs": 3, "tol_exp": 6, "sec": 0.77} diff --git a/problems/crouzeix/data/certify.log b/problems/crouzeix/data/certify.log new file mode 100644 index 0000000..5f2b7b5 --- /dev/null +++ b/problems/crouzeix/data/certify.log @@ -0,0 +1,70 @@ + 25/286 certified, 21s elapsed + 50/286 certified, 40s elapsed + 75/286 certified, 60s elapsed + 100/286 certified, 78s elapsed + 125/286 certified, 96s elapsed + 150/286 certified, 114s elapsed + 175/286 certified, 132s elapsed + 200/286 certified, 150s elapsed + 225/286 certified, 168s elapsed + 250/286 certified, 186s elapsed + 275/286 certified, 206s elapsed +certified 286 points, refutations: 0, total 214s + +independent verification (verify_ratio.py) of the top 8 endpoints by float ratio: +[jordan2 seed=602007 float=1.99875] + characteristic polynomial agrees with the determinant: OK + ||p(A)||^2 definiteness [1.708227933, 1.708228853] roots [1.708227360, 1.708228007] AGREE + 32 support values, 0 disagreeing with the reference OK + outer bound on max|p|^2 reference 0.442922732 independent 0.442922789 AGREE + ratio reference [1.963852766, 2.003138920] independent [1.963852309, 2.003138424] AGREE + verify: AGREE +[real seed=406009 float=1.99864] + characteristic polynomial agrees with the determinant: OK + ||p(A)||^2 definiteness [3.600167515, 3.600168040] roots [3.600167714, 3.600168270] AGREE + 32 support values, 0 disagreeing with the reference OK + outer bound on max|p|^2 reference 0.927922296 independent 0.927922221 AGREE + ratio reference [1.969724783, 2.051433206] independent [1.969724918, 2.051433272] AGREE + verify: AGREE +[real seed=506001 float=1.99863] + characteristic polynomial agrees with the determinant: OK + ||p(A)||^2 definiteness [12.270158291, 12.270159108] roots [12.270158009, 12.270158838] AGREE + 32 support values, 0 disagreeing with the reference OK + outer bound on max|p|^2 reference 3.200673941 independent 3.200673858 AGREE + ratio reference [1.957962482, 2.047418460] independent [1.957962485, 2.047418438] AGREE + verify: AGREE +[peaks seed=605007 float=1.99836] + characteristic polynomial agrees with the determinant: OK + ||p(A)||^2 definiteness [739.814318590, 739.814319368] roots [739.814318093, 739.814318783] AGREE + 32 support values, 0 disagreeing with the reference OK + outer bound on max|p|^2 reference 199.400872260 independent 199.400823580 AGREE + ratio reference [1.926184303, 2.398949702] independent [1.926184538, 2.398949701] AGREE + verify: AGREE +[real seed=406007 float=1.99818] + characteristic polynomial agrees with the determinant: OK + ||p(A)||^2 definiteness [2.305948178, 2.305948783] roots [2.305948359, 2.305949160] AGREE + 32 support values, 0 disagreeing with the reference OK + outer bound on max|p|^2 reference 0.597578871 independent 0.597579129 AGREE + ratio reference [1.964387468, 2.023830686] independent [1.964387121, 2.023830851] AGREE + verify: AGREE +[jordan3 seed=601006 float=1.99677] + characteristic polynomial agrees with the determinant: OK + ||p(A)||^2 definiteness [15.012566390, 15.012566948] roots [15.012566509, 15.012567464] AGREE + 32 support values, 0 disagreeing with the reference OK + outer bound on max|p|^2 reference 4.047140827 independent 4.047140881 AGREE + ratio reference [1.925986826, 2.072421238] independent [1.925986821, 2.072421274] AGREE + verify: AGREE +[jordan2 seed=402005 float=1.99675] + characteristic polynomial agrees with the determinant: OK + ||p(A)||^2 definiteness [0.847132360, 0.847133323] roots [0.847132185, 0.847133124] AGREE + 32 support values, 0 disagreeing with the reference OK + outer bound on max|p|^2 reference 0.220015754 independent 0.220016015 AGREE + ratio reference [1.962224739, 2.022964281] independent [1.962223371, 2.022964043] AGREE + verify: AGREE +[jordan2 seed=602004 float=1.99669] + characteristic polynomial agrees with the determinant: OK + ||p(A)||^2 definiteness [1.436166928, 1.436167773] roots [1.436167124, 1.436167710] AGREE + 32 support values, 0 disagreeing with the reference OK + outer bound on max|p|^2 reference 0.371746448 independent 0.371746011 AGREE + ratio reference [1.965527149, 2.047954503] independent [1.965528439, 2.047954458] AGREE + verify: AGREE diff --git a/problems/crouzeix/data/classified.jsonl b/problems/crouzeix/data/classified.jsonl new file mode 100644 index 0000000..447f2bc --- /dev/null +++ b/problems/crouzeix/data/classified.jsonl @@ -0,0 +1,286 @@ +{"family": "random", "seed": 400000, "x": [1.1637546894699131, 1.3387571672177243, 1.6092230250910644, -0.6324386977805513, -0.5786510560275908, 1.3807163847105475, -1.2770144373189805, 0.3327930226284563, 0.6855809351172977, 0.6567599551279006, 0.8653677728707757, 1.6492690818808908, -0.4680017551301743, 1.9717084498434407, 1.0749568514281624, -1.1195807394928647, -1.1194896571507935, 0.541507939218701, 1.5546265891994846, -0.9748372026574621], "ratio96": 1.4733466106055264, "ratio_final": 1.474585189086896, "diag": {"ratio": 1.474585189086896, "nonnormal": 0.6815660377178587, "eig_gaps": [0.45026467441734236, 0.5220817275692271, 0.7680839058690033], "discness": 0.23395279665038896, "peaks": 3, "deg": 3, "diamW": 3.8708585980884043}, "class": "stall", "probe_best": 1.495530747691867} +{"family": "jordan3", "seed": 401000, "x": [0.16813274325574726, 0.1267920593445052, -0.2269579418812387, 0.29234757829988345, -0.45320988643118, 0.2065849867685672, 1.9578803040341368, -0.38386813569009537, -0.42811876624542855, 0.026407176195512028, 2.14531753556515, -0.039140190899580526, -0.12054782373550071, -0.21518652393552956, 0.1798048834730397, -1.617509404769688, 2.5115610147398932, -1.4266432013885009, 0.29729074082271817, -0.12175554302440633], "ratio96": 1.973242533081747, "ratio_final": 1.974033911016823, "diag": {"ratio": 1.974033911016823, "nonnormal": 0.7490028296412479, "eig_gaps": [0.08074737656884254, 0.14295757990216038, 0.2090576114453739], "discness": 0.005388684247591025, "peaks": 3, "deg": 3, "diamW": 2.9965194024681}, "class": "near2-double", "probe_best": 1.974033911016823} +{"family": "jordan2", "seed": 402000, "x": [0.2118205430836373, -0.12011191791608411, -0.1771870327824911, 0.02853799878724763, 0.1953939747827479, 0.34748575426978234, 1.4142097830310956, -0.29499659515509274, 0.33319460297083336, 0.36328377180905314, -0.028813029677675572, -0.0007975654446331134, -0.08344206619356137, -0.02209705102198633, 1.739516935276686, -0.49861528316727655, -0.020244626630011864, -0.14025816300462812, -0.03301056786922761, 0.01001330726583325], "ratio96": 1.9724002360395447, "ratio_final": 1.972475733235554, "diag": {"ratio": 1.972475733235554, "nonnormal": 1.3255588791027426, "eig_gaps": [0.2630065212921424, 0.29549658029597026, 0.30974893270759923], "discness": 0.006665045235292756, "peaks": 3, "deg": 3, "diamW": 1.583389266309631}, "class": "near2-other", "probe_best": 1.972475733235554} +{"family": "nearnormal", "seed": 403000, "x": [-0.8340652029109421, 1.3776805332489246, -0.89902307819048, 0.003539677027255185, -0.14904558094480436, 1.1048924801555073, -0.9424538495244978, -0.31354668908244854, -1.4839312764248733, 0.18165676074196407, 0.14732872225860344, 0.9197300442040808, 2.945779862567481, 1.4171345279649459, 2.8282302783765125, -0.8648540507100315, 3.6519454847211708, 0.7840822699605925, -0.06953594025205653, 0.5693541951192229], "ratio96": 1.636683224465937, "ratio_final": 1.6399400334716803, "diag": {"ratio": 1.6399400334716803, "nonnormal": 0.5800868283122488, "eig_gaps": [0.2974564409894917, 0.5375404212731334, 0.5549751764635916], "discness": 76248354153.57133, "peaks": 3, "deg": 3, "diamW": 2.4788051611325033}, "class": "stall", "probe_best": 1.728109072705156} +{"family": "companion", "seed": 404000, "x": [-0.302070942800376, 0.7186957327412735, -1.220431966355091, -0.28302913538871954, 1.1922976298060095, -0.3861345695243055, 0.006742010288531117, 0.9868143418417582, -2.9363934560295832, -0.8461785373454943, 1.4725339643484996, -0.8117568349307664, 0.24565062629463483, -0.19163772035101503, 0.3483504550341804, 1.8304935890793452, 0.10378057082829252, -0.018477927634751905, -0.0024398583052655607, -0.032012284175236785], "ratio96": 1.9233604487589435, "ratio_final": 1.925728641730908, "diag": {"ratio": 1.925728641730908, "nonnormal": 1.3117474161398588, "eig_gaps": [0.31479708115845295, 0.4304908255783476, 0.5593980397743075], "discness": 0.047766850371888075, "peaks": 4, "deg": 3, "diamW": 4.317018420032665}, "class": "stall", "probe_best": 1.9865016767124348} +{"family": "peaks", "seed": 405000, "x": [-0.9829424448880801, -0.7162593769905634, -0.24827329725911917, 0.10670932013826717, 0.8170578187790367, -0.7850402123191147, 1.6134320656710042, 1.0544961635406118, 0.18325311319467358, -0.10721074717367185, 1.7917002484353386, -0.07948521792676548, -0.6293219268544203, -1.6296960493545187, -1.8190271942750518, 0.6733837339392079, 0.0859049105321592, 1.6120233534071011, -0.4721426411982612, 0.14468568082262584], "ratio96": 1.9591307275493353, "ratio_final": 1.9590367342496156, "diag": {"ratio": 1.9590367342496156, "nonnormal": 0.7301317237937558, "eig_gaps": [0.343845566856024, 0.4330225430648908, 0.5614416708015036], "discness": 0.07597811246029411, "peaks": 5, "deg": 3, "diamW": 3.2083722848565004}, "class": "stall", "probe_best": 1.9596108825163632} +{"family": "real", "seed": 406000, "x": [1.5230879030064817, 0.1917667455752054, 1.1691083250469634, 0.040517370157498095, 1.2261733719182404, 0.03094809369655139, -0.0578445295265749, 0.30359123280601236, -0.7908814716316774, -0.30150336530603894, -0.14269043710767776, 0.008289110011652815, -1.5339787728821053, 0.10868625788913289, -0.6099159703026269, 2.892241883098684e-05, 1.6065284472280585, -0.49408933258631627, -0.2823879295699706, 0.23888531386791997], "ratio96": 1.9504599929031357, "ratio_final": 1.9563442656411278, "diag": {"ratio": 1.9563442656411278, "nonnormal": 0.20410747695370537, "eig_gaps": [0.059857837384671014, 0.3493181045882263, 0.3982177786335099], "discness": 9790131382.9728, "peaks": 3, "deg": 3, "diamW": 0.9666540940055174}, "class": "stall", "probe_best": 1.974242773834062} +{"family": "random", "seed": 400001, "x": [1.2534022026553093, -0.7839057835795689, -0.6320330927185923, 1.163510247712968, 0.4654765054130657, -0.7511230684899993, 0.7289954445143618, 1.9911578814058042, -1.871329088365571, 0.4283283932361738, 2.200024191893569, 0.6452098054736204, -1.115536301921391, -4.782942911108577, 2.771124032085873, -0.8636181430322365, -4.048686985588349, -0.1899531405733032, -0.21753776402531297, 0.44097331132699996], "ratio96": 1.9495687744872776, "ratio_final": 1.9495289236168243, "diag": {"ratio": 1.9495289236168243, "nonnormal": 0.741336075881184, "eig_gaps": [0.16960800350709668, 0.4746412734077458, 0.5829740901127468], "discness": 0.10784089552554417, "peaks": 3, "deg": 3, "diamW": 4.649588289840303}, "class": "stall", "probe_best": 1.9503641290319245} +{"family": "jordan3", "seed": 401001, "x": [0.459123380682876, 0.26786964995147866, 0.11201371160172235, 0.0886819681920511, -0.4774684099526276, 0.3787335545823378, 1.251189816791397, -0.4034200451877263, -0.03984505200977382, 0.11594737599622738, 1.1883934669823124, 0.22274549982880473, -0.31537636331976127, 0.04990318284169133, -0.02939474074717155, -0.7597996794741645, 1.316283271329167, 0.11335949855014238, -0.019377717460602698, -0.31660698732469605], "ratio96": 1.9895992576924333, "ratio_final": 1.989554662554931, "diag": {"ratio": 1.989554662554931, "nonnormal": 0.7034575313536575, "eig_gaps": [0.19239505283938285, 0.3235760914393243, 0.46451297853252504], "discness": 0.055174570508538585, "peaks": 4, "deg": 3, "diamW": 2.0303639786076593}, "class": "near2-other", "probe_best": 1.989554662554931} +{"family": "jordan2", "seed": 402001, "x": [-0.7025273628037407, -0.4397797168352926, 0.2532139695733695, 0.01625216637404006, 0.290104253732692, -0.16539956349199142, 1.6977196047063063, -0.5864427883957316, 0.22484896447734037, -0.8462954944338525, 0.058045183643668896, -0.0881592432452249, 0.28337203601165384, 0.023275679841948796, 0.6935824464036187, -0.6038792672100404, -0.05588925138731571, 0.031973418086524555, -0.004829612918725079, 0.051584593890371844], "ratio96": 1.9694414648093004, "ratio_final": 1.969392851389826, "diag": {"ratio": 1.969392851389826, "nonnormal": 1.3207392998078549, "eig_gaps": [0.08142208947786002, 0.4523795695427998, 0.4651663729373806], "discness": 0.061776182893447475, "peaks": 4, "deg": 3, "diamW": 2.276529260841256}, "class": "stall", "probe_best": 1.975697860288861} +{"family": "nearnormal", "seed": 403001, "x": [-0.009475797701735617, 0.3974203276399666, -0.4940486298404265, 1.024892955655842, 0.12297324047548269, 1.6261423208733339, 0.3998123904387274, 0.3831409906146256, -1.353447570891378, -0.2885291068378949, 0.5582969342576793, -0.5075173362106897, 6.082011666918385, 6.989655394010462, -9.945138419198493, 5.250400975088873, 1.0624609868592616, -4.551583688212414, 0.838010192856808, -0.08642697386836584], "ratio96": 1.6147835185003712, "ratio_final": 1.667019965562423, "diag": {"ratio": 1.667019965562423, "nonnormal": 0.7115820963332398, "eig_gaps": [0.3820243680677492, 0.4151380147068706, 0.5955104162876467], "discness": 52798461394.470085, "peaks": 2, "deg": 3, "diamW": 2.075261754638286}, "class": "stall", "probe_best": 1.6747690688600894} +{"family": "companion", "seed": 404001, "x": [0.559644580272846, 1.2092786314653692, 0.34640786001761215, -1.3134996879945944, -0.01278807055400431, -1.0386209380715925, 2.2028205130881546, -2.4049062555848604, 0.4362603005498601, -1.6093874924190836, 1.6876497670606865, 1.9354975281553213, 0.29433088194514895, -0.29237929129913853, 0.019727497321109115, 1.0135081101970773, 0.7666431539448728, -0.619347057000355, 0.007807150569793326, 0.12466181742766147], "ratio96": 1.8610581695474009, "ratio_final": 1.9249874758765917, "diag": {"ratio": 1.9249874758765917, "nonnormal": 0.8085260109936573, "eig_gaps": [0.0878073052446734, 0.45031838777416466, 0.49150051468002887], "discness": 0.05929225666067442, "peaks": 3, "deg": 3, "diamW": 5.151111900510527}, "class": "stall", "probe_best": 1.9704819274299838} +{"family": "peaks", "seed": 405001, "x": [-0.02173524795549614, -0.39601778332558846, 0.18692974993208444, 0.5299692118620043, 0.4585859149183178, 0.02576874664711823, 0.988407888166897, 2.3552294155351343, -0.5548415272229308, -0.1488470363138631, 2.5064878763822014, -0.3423311152537711, -1.3123293255189705, 0.2653458130092346, 3.293248397562227, 0.7199416854723517, -7.314313228557157, 0.7841829763824721, -0.5179216052564788, 0.05726850429804414], "ratio96": 1.988013301339618, "ratio_final": 1.987997920438502, "diag": {"ratio": 1.987997920438502, "nonnormal": 0.727576425892596, "eig_gaps": [0.15222560481144842, 0.16990120226686875, 0.2522908173898413], "discness": 0.011653711960387395, "peaks": 3, "deg": 3, "diamW": 3.7623506029783775}, "class": "near2-other", "probe_best": 1.987997920438502} +{"family": "real", "seed": 406001, "x": [-0.988113351149956, -0.004215695963268935, -0.10783234770594591, -0.014712474627010544, -0.74256409378083, -0.05630888044065854, -1.2366422453867592, -0.1867344033911941, -0.2210236323594455, 0.02577977929101017, -1.3427010258243375, 0.10384361785827825, 0.31876850562860204, 0.027435905379162365, 1.059872620387306, 0.054812150354345585, 0.5836337044098387, -0.032548936295118014, -0.22242323129668862, -0.03748991310826398], "ratio96": 1.987049615233899, "ratio_final": 1.987039203566344, "diag": {"ratio": 1.987039203566344, "nonnormal": 0.5737179953902536, "eig_gaps": [0.12258409862794858, 0.31063945689876693, 0.42992036733957695], "discness": 0.08006903626634833, "peaks": 3, "deg": 3, "diamW": 2.04768987870431}, "class": "near2-double", "probe_best": 1.987039203566344} +{"family": "random", "seed": 400002, "x": [2.2965814774339073, 3.2318532328058307, 1.481088281830253, -2.7417876069041736, -1.9397073521240773, -0.06358556137594344, 1.1101119646631696, -5.307475412800426, 4.506333248459718, 0.7605650059422844, -4.396211518723932, -0.6559125604562352, 3.1606539288619127, -0.6530344188555036, 1.3958658774037316, 1.8692336401755858, -1.7928349216148893, -2.8902811826895602, 0.21989494617164101, -0.005988294238791124], "ratio96": 1.8845034327679335, "ratio_final": 1.8855779399806656, "diag": {"ratio": 1.8855779399806656, "nonnormal": 0.8095592008327809, "eig_gaps": [0.429862475460481, 0.5310464177097892, 0.5965395757537864], "discness": 0.04318252542616364, "peaks": 4, "deg": 3, "diamW": 10.106701664053938}, "class": "stall", "probe_best": 1.8982121219281036} +{"family": "jordan3", "seed": 401002, "x": [0.20139600696814852, 1.064168733153285, -0.35321485651948403, -0.34082059765571016, -0.28413044129278153, -1.0680420728159072, 0.33249151280454625, -0.026997359810870012, 0.3338861048819237, 0.045460280544681014, 1.1809106299276404, -0.09973370120148352, 0.589510188668531, -0.0727329591203819, 0.10951727492730978, -0.1279616031082924, 0.8368803277298831, 0.4908092341074186, 0.16733697682434046, -0.24588480582936134], "ratio96": 1.7687693957512514, "ratio_final": 1.7552977507389464, "diag": {"ratio": 1.7552977507389464, "nonnormal": 0.6315317711148664, "eig_gaps": [0.28391052032307385, 0.587060083806575, 0.8499069423502256], "discness": 0.32449348442431614, "peaks": 4, "deg": 3, "diamW": 2.5729780520485184}, "class": "stall", "probe_best": 1.8360492073231853} +{"family": "jordan2", "seed": 402002, "x": [-0.17361558324087356, 0.13403392790124413, -0.016709072423486196, -0.215016720707923, -0.1611948185527029, 0.3040663167515961, 1.2358625671654415, -0.059345765614480156, -0.5097572326442035, -0.022739466242883907, -0.11796971001040496, 0.24899904588950472, 0.14305019883988235, 0.10641089683623506, 1.7472768250891861, 0.19197298585687284, 0.01907917764283925, 0.02010315217455351, 0.04530198215927966, -0.03748807907316161], "ratio96": 1.9953873588599265, "ratio_final": 1.995382827290676, "diag": {"ratio": 1.995382827290676, "nonnormal": 1.3096591973131966, "eig_gaps": [0.12080187113680109, 0.27116893648586154, 0.3817924270528878], "discness": 0.013093141115402059, "peaks": 3, "deg": 3, "diamW": 1.4112815337368492}, "class": "near2-double", "probe_best": 1.995382827290676} +{"family": "nearnormal", "seed": 403002, "x": [1.098821609310131, -0.9408546793452492, 0.8975718933055052, -0.9767392254631577, 1.1196559334782767, -1.2100051294064582, 0.09724297846315663, 0.05772732859267432, -0.09548194490342883, -0.2589530578067273, 0.697017583581543, 0.04679472355237651, -0.4097027358661447, 1.2392651365918743, -0.9641392252067646, -4.6365414160419345, -2.376110461528448, 3.988833144900621, 1.1138723542647768, -0.0040385892590619425], "ratio96": 1.8933692608818016, "ratio_final": 1.8941728389597947, "diag": {"ratio": 1.8941728389597947, "nonnormal": 0.12372249642697516, "eig_gaps": [0.24712456054766652, 0.32634466071097823, 0.38935463006481746], "discness": 10553153873.514143, "peaks": 3, "deg": 3, "diamW": 0.8272101463846251}, "class": "stall", "probe_best": 1.9326709708230159} +{"family": "companion", "seed": 404002, "x": [0.16594654557402855, 0.1325588312465653, 0.08509788150556089, -0.1737939455064732, 1.9838025032853912, 1.3296869962587827, -3.450607048506513, 1.7169113393229505, -0.48910795872714563, 0.07336391694463582, 3.0577482792707453, 1.2164242578411335, -0.33213021260968756, 0.09341151192922283, 2.3477713166480587, -5.481605807864188, 0.14966305432022564, 2.775477680130379, 0.1908739794371842, 0.07287145472301515], "ratio96": 1.9711905081361913, "ratio_final": 1.9756485744848153, "diag": {"ratio": 1.9756485744848153, "nonnormal": 0.6756619406061208, "eig_gaps": [0.05524106542609626, 0.3794936520502056, 0.4222540104997716], "discness": 0.05927475397182865, "peaks": 3, "deg": 3, "diamW": 5.735615133008684}, "class": "near2-double", "probe_best": 1.9756485744848153} +{"family": "peaks", "seed": 405002, "x": [-1.2357904607336154, -0.02652099922100296, 0.47381033338597156, 0.7246994985785067, 0.7320243321857363, 1.2767440073004872, 1.1724237115692469, -0.1264005301237222, -0.1372222266740568, -0.033456219283891, 1.2764445365839505, 1.4305699752382663, -3.0001061066387638, -0.7337219395312634, -1.5029442367895651, -3.2097927065579706, 2.7760880927291636, -1.6752083917275167, -0.22999947522037434, 0.40491075354533634], "ratio96": 1.8883416000820648, "ratio_final": 1.8892524871320382, "diag": {"ratio": 1.8892524871320382, "nonnormal": 0.5933349916317641, "eig_gaps": [0.1819273360376797, 0.5574308777553952, 0.7045623749860311], "discness": 0.27539248149602147, "peaks": 4, "deg": 3, "diamW": 3.3499562172964015}, "class": "stall", "probe_best": 1.9116418865647458} +{"family": "real", "seed": 406002, "x": [2.57330113920448, 1.0470074709203887, -3.561689272740095, 1.340685173213159, 2.705031382423333, 1.623867511292111, 2.906300750450799, 1.9280797358702975, -5.782088311032604, -0.6494012370289151, -3.650467216799422, -1.0218439336679426, 4.307155666179382, -3.950286431639208, 2.2948510668011064, 2.7714256911911246, 2.4767759274068704, 0.14123535095861245, -0.16064770571269726, 0.08360002206249093], "ratio96": 1.5307624397625792, "ratio_final": 1.5424697871862874, "diag": {"ratio": 1.5424697871862874, "nonnormal": 0.8027542317247043, "eig_gaps": [0.05630620571677072, 0.584464964993525, 0.5969402622186633], "discness": 0.227202665962716, "peaks": 3, "deg": 3, "diamW": 10.508782956814636}, "class": "stall", "probe_best": 1.704314932785983} +{"family": "random", "seed": 400003, "x": [-4.143666743522802, -2.4372769532346457, 2.599104236417321, -2.5585472364811226, 3.0784887547565933, 2.9583409434589676, 2.0384036611578313, -0.38927419147899145, -0.9000404716592671, -2.8228240095436874, 1.7280353191665332, 1.0218940294326686, -1.734139221990316, 2.1333807003688396, 0.6053124842639921, -3.658212048668629, -0.6579889143987272, -0.8791447075668719, 0.12500420219225294, -0.1637261875218362], "ratio96": 1.3338882421899658, "ratio_final": 1.3693221045958268, "diag": {"ratio": 1.3693221045958268, "nonnormal": 0.6444518263110308, "eig_gaps": [0.5577616311774807, 0.6792500292747193, 0.9080130622086073], "discness": 0.19113494168280235, "peaks": 3, "deg": 3, "diamW": 9.928393283842611}, "class": "stall", "probe_best": 1.369974138933139} +{"family": "jordan3", "seed": 401003, "x": [-0.3238977360555537, -0.015273286900158209, 0.7679847592702946, 0.39741178137707, -0.33366993800244116, -0.42840451207797126, 1.7695251654058153, 0.359136056484542, -0.5196049561163365, 0.0731330642687805, 1.6202210158073578, 0.7468723872367691, -0.023356757887338632, -0.44829361607431006, 0.07156672719085405, 0.2524214058472434, 2.1772446915116728, 0.47562544413736463, -0.310598264286569, -0.0082592132432188], "ratio96": 1.9644149141611291, "ratio_final": 1.9646022916763135, "diag": {"ratio": 1.9646022916763135, "nonnormal": 0.725743885901117, "eig_gaps": [0.1392445027340264, 0.39331402806004195, 0.4639206613911675], "discness": 0.049055151561915825, "peaks": 3, "deg": 3, "diamW": 2.967778093112101}, "class": "stall", "probe_best": 1.9651934388397856} +{"family": "jordan2", "seed": 402003, "x": [0.29820223367825005, -0.008916112291766053, -0.11147867905317213, -0.025942098814447076, 0.3211006336340839, -0.0008192199266728213, 0.8464313548479238, 0.3946673495363116, 0.6786352120706169, 0.23592157168362995, -0.37342309988386885, 0.11048073457228026, -0.1509486910263062, -0.026962800381941588, 1.9153872639476635, 0.3789983778752231, 0.004024150203612121, 0.015252897226587066, -0.11790192348897544, 0.018846940685068283], "ratio96": 1.9942133187180406, "ratio_final": 1.9942564539334426, "diag": {"ratio": 1.9942564539334426, "nonnormal": 1.3012771164566952, "eig_gaps": [0.01866508222205502, 0.31511022429902275, 0.3329959685578438], "discness": 0.02018950040083996, "peaks": 4, "deg": 3, "diamW": 1.3012416645387301}, "class": "near2-double", "probe_best": 1.9942564539334426} +{"family": "nearnormal", "seed": 403003, "x": [0.612775113115535, -0.08456720812126954, 0.410604527368326, -0.7011073493200497, 1.1737415067311219, 1.2141837847068544, 0.5985521648646562, 0.9586704491494213, -0.051205778465194474, -0.9226335848585832, 1.0247151843479783, -0.1991496220799942, 0.9757529384266178, -1.0640288768078818, -1.3833800631261606, 0.5269437524110984, 0.0633362229041409, -1.0569570053578814, 0.7032585665808946, 0.026458805866413693], "ratio96": 1.6710239962764462, "ratio_final": 1.7126386013132484, "diag": {"ratio": 1.7126386013132484, "nonnormal": 0.663967925161178, "eig_gaps": [0.2350664768114006, 0.5125348464667521, 0.746936262867808], "discness": 0.4107229355255806, "peaks": 3, "deg": 3, "diamW": 2.760244871668558}, "class": "stall", "probe_best": 1.7186486256923375} +{"family": "companion", "seed": 404003, "x": [-0.6191856827551976, 1.3885205312751767, -0.05488967841986589, 1.2534901237661673, -0.48665806964423997, -0.7145381435425232, 1.0574906314230224, -0.232345917608567, 0.45362983778029375, 0.6297498935215046, 1.3343000711559556, -0.15531167290919334, -0.012031117033858826, 0.14395729658669987, 0.054564335667510594, 0.42637899247333666, -0.06807910656527888, -0.01642576497166228, 0.18787057970143517, 0.09863440923388167], "ratio96": 1.6762682993505984, "ratio_final": 1.6764003782028387, "diag": {"ratio": 1.6764003782028387, "nonnormal": 0.6337571682462164, "eig_gaps": [0.19819659032163844, 0.6882366572643926, 0.7197975339700344], "discness": 0.2508189465341587, "peaks": 4, "deg": 3, "diamW": 2.927531921149381}, "class": "stall", "probe_best": 1.6818312609974597} +{"family": "peaks", "seed": 405003, "x": [0.06488670547085905, 0.6571318627495162, -0.0719574579837327, 0.6868121035437811, 0.23314002127388653, 0.9519772817956196, 0.031624631256464336, 0.011265243388016448, -0.24202815946953418, -0.14634732226761502, 0.9979310789869189, -0.3292474460851618, -0.26075654194429626, 2.310178598556929, -2.4132684724596905, -0.0008316773648728481, -0.9222523682132648, 0.013481874374360785, 0.12310139346277692, -0.18251964856455416], "ratio96": 1.9708704792724532, "ratio_final": 1.9712314269731561, "diag": {"ratio": 1.9712314269731561, "nonnormal": 0.5841417624508887, "eig_gaps": [0.1206200175609261, 0.2924276350145401, 0.348203580562967], "discness": 12760030514.373123, "peaks": 3, "deg": 3, "diamW": 1.1608841043969915}, "class": "near2-double", "probe_best": 1.9712314269731561} +{"family": "real", "seed": 406003, "x": [-0.36948292612956335, -0.1286238434820614, -0.24586482803610799, -0.08388095442261244, -0.26245537420859466, -0.023813641866055585, -0.03414374399308741, 0.22665662039408874, -0.6715005291132485, 0.270489651814715, 0.4013567471189161, 0.23392967975543555, -0.5451519702980452, -0.032033646760559684, -1.7609553833314866, 0.4621121664432868, 0.20409457512107254, -0.07943699514189187, -0.08347471948523372, -0.1421523762464439], "ratio96": 1.9821370321564569, "ratio_final": 1.9829369879028436, "diag": {"ratio": 1.9829369879028436, "nonnormal": 1.03491657023867, "eig_gaps": [0.06900230132034489, 0.14557125140392385, 0.16587222792594403], "discness": 0.021745888893262677, "peaks": 3, "deg": 3, "diamW": 0.9031054774697571}, "class": "near2-double", "probe_best": 1.9829369879028436} +{"family": "random", "seed": 400004, "x": [-1.0238424224522593, -0.19073810199422947, 0.38405792692029905, -0.8022105541133202, -1.1824458412509555, -1.4664708502019321, -0.38013318900792903, 1.5993998908247278, 0.41810270410820666, 0.7244130742447255, 0.06938873745905763, 1.1685220515834667, 2.214099374005669, 1.2490767833295962, 2.97570522561518, -1.8890249076990755, 1.1799754611713338, 0.3781940080305125, 1.3684438531240062, 0.36489145162399605], "ratio96": 1.6736653118177551, "ratio_final": 1.6791706927654377, "diag": {"ratio": 1.6791706927654377, "nonnormal": 0.5741465651442422, "eig_gaps": [0.4641809136863659, 0.5542327068120335, 0.6143762580333431], "discness": 0.15910230762251326, "peaks": 3, "deg": 3, "diamW": 2.769510686511074}, "class": "stall", "probe_best": 1.7351664745654032} +{"family": "jordan3", "seed": 401004, "x": [-0.04353481431211606, 0.10555704414719758, 0.2568938143382402, 0.3658538585599047, -0.25896148447226996, -0.17881485696830512, 0.8741741170680681, -0.3553568960972869, -0.34210399339251807, 0.3930810307242405, 0.8677819627013804, 0.35981953697070346, -0.02732395088410726, -0.07481517742477534, -0.036059169539717426, -0.3521367747310453, 1.3646254433352318, -0.07179681221346128, 0.20009398981808957, -0.2933203248633396], "ratio96": 1.9913094441734593, "ratio_final": 1.9912826371971442, "diag": {"ratio": 1.9912826371971442, "nonnormal": 0.7199149575647555, "eig_gaps": [0.22182487078105834, 0.24716198801164574, 0.466447391374071], "discness": 0.04583853696790835, "peaks": 4, "deg": 3, "diamW": 1.608285217490507}, "class": "near2-other", "probe_best": 1.9912826371971442} +{"family": "jordan2", "seed": 402004, "x": [-0.06613300018958321, -0.04404276550997127, -0.2665646803887126, -0.47296028699590464, -0.0017192160906624497, 0.18808928986884255, 0.42691792463692907, -1.432704556380961, 0.73189821338082, 0.16160792587049322, 0.2905947714996434, -0.26281152042302935, 0.25038327725847065, 0.08635932795656015, 0.6273659131165013, -0.5273341440183246, 0.05863062251495771, 0.019253481824348717, 0.022355484544824222, -0.003350326751671481], "ratio96": 1.9910819828899966, "ratio_final": 1.99108154963197, "diag": {"ratio": 1.99108154963197, "nonnormal": 1.3116169295423288, "eig_gaps": [0.13484187675173773, 0.2649992378570075, 0.3986038105723817], "discness": 0.022120510597794994, "peaks": 4, "deg": 3, "diamW": 1.7865618705069537}, "class": "near2-double", "probe_best": 1.99108154963197} +{"family": "nearnormal", "seed": 403004, "x": [-0.6411055993450849, -0.5833352267880088, 0.6617136281014167, -0.7976375265777604, 0.06855979038471052, 2.0122028391320224, 0.22852030800136983, 0.09863030345176402, 8.656574995424773e-06, 9.070714957836103e-06, 1.4564311662302722e-06, 1.0320777078853687e-05, 0.16897081073411543, -0.1747762849859956, -0.15270078965358894, -0.21530395805557664, 0.17654620631582502, 0.5469461325493061, 1.006524511384689, 0.1133941856499105], "ratio96": 0.99999999996982, "ratio_final": 0.9999999999820248, "diag": {"ratio": 0.9999999999820248, "nonnormal": 0.07960469144573055, "eig_gaps": [0.4547942128992848, 0.9268639393750735, 0.9891959366116936], "discness": 0.48054407625096046, "peaks": 1, "deg": 3, "diamW": 2.9031306318763153}, "class": "one-exact", "probe_best": 0.9999999999834525} +{"family": "companion", "seed": 404004, "x": [-1.374188638444159, -1.4800292498941063, 0.7468317690637432, 1.1795774726458563, 0.5203696347398206, -0.48580480017974326, 3.128797755130753, 3.3302014643679474, -3.1883551684073073, 0.43717988555093124, 4.446566712056727, 0.06526209498533297, 0.4497793503392589, -0.4412636131613182, -1.5833072302981062, -0.6829246160014771, -5.086025595072723, 8.928486442249206, 0.38878193718169407, 0.7143814827551975], "ratio96": 1.9209336267040593, "ratio_final": 1.9219565692814136, "diag": {"ratio": 1.9219565692814136, "nonnormal": 0.8032667532939068, "eig_gaps": [0.2163069879970563, 0.27536451595623446, 0.4378109500516135], "discness": 0.032662213097169795, "peaks": 3, "deg": 3, "diamW": 7.770017767143024}, "class": "stall", "probe_best": 1.9253550889532653} +{"family": "peaks", "seed": 405004, "x": [-1.1623865848615051, 0.6509958491850696, -0.5003360230842877, -0.020101183726879884, -0.5036340490734614, -0.03659664609401815, 1.295354048316115, 1.185393715540028, 1.2211443077978765, 1.164968203505442, 0.1356042721765338, 0.2792774311092724, -2.8360542513331763, 2.346435049043228, -3.85639244226409, 0.9658545591901251, 0.22344365455954474, 0.21076864046464272, 0.0497471499238199, 0.08357041257100828], "ratio96": 1.9863456648712265, "ratio_final": 1.9863425155864072, "diag": {"ratio": 1.9863425155864072, "nonnormal": 1.0798935920298736, "eig_gaps": [0.006412076312312546, 0.35933221353351663, 0.3629642979288813], "discness": 0.08568097426039863, "peaks": 4, "deg": 3, "diamW": 2.623475915997666}, "class": "near2-double", "probe_best": 1.9863425155864072} +{"family": "real", "seed": 406004, "x": [-0.6833724558009195, 0.28590789581862186, 0.8368358877651443, -0.3257315338483946, 0.10432383941304829, -0.5299022869252201, 2.3320326289319167, 0.022248129832376513, -1.3241970803462717, -0.23767502542305508, 2.2000821555742833, 1.0222498764248833, 0.383337091724671, -0.6626213822858326, 0.37451269623021455, -0.715716681206092, -1.9627507185007529, -0.21778252830792333, -0.1822785937833571, 0.1253179709716135], "ratio96": 1.9585816743666855, "ratio_final": 1.9588423767255672, "diag": {"ratio": 1.9588423767255672, "nonnormal": 0.7735703190313216, "eig_gaps": [0.19145450636405284, 0.28551348787183506, 0.4125601991633988], "discness": 0.039589627671257795, "peaks": 3, "deg": 3, "diamW": 3.9718773732626405}, "class": "below2-max", "probe_best": 1.9589670938233057} +{"family": "random", "seed": 400005, "x": [-1.480230380410387, -2.7358386026723704, -0.08284297081073352, 0.9112889916367062, 1.1216636038828318, -0.4256150714809218, -2.888519132210435e-05, 3.107237571685124e-06, -1.861481800006378e-06, -7.114232940708941e-06, -0.17769902164069284, 0.3878543746980968, 1.1388411790919544, -0.07150931169589188, 0.10535510326645858, -0.5890000408290148, -0.8217765896373872, -0.7172422981592165, 1.1649345835032374, -0.695842962736767], "ratio96": 0.9999999999664926, "ratio_final": 0.9999999999752113, "diag": {"ratio": 0.9999999999752113, "nonnormal": 0.09195209432355816, "eig_gaps": [0.454513338269815, 0.878853023554804, 0.9864916438477316], "discness": 0.8940725258068817, "peaks": 1, "deg": 3, "diamW": 3.959148141072227}, "class": "one-exact", "probe_best": 0.9999999999754693} +{"family": "jordan3", "seed": 401005, "x": [0.08780466826470232, 0.13202629980266084, -0.5378595702294608, -0.1800878564795924, 0.10156906288870116, 0.2933880689784183, 1.2485115814755958, 0.5899745534209925, 0.6888473452569717, 0.1671128388945428, 1.3872332832468959, 0.2411801182270396, 0.04407989151121971, -0.15559033790406324, 0.4600510416361281, -0.1343684200355708, 1.3702500874669106, 0.29762339978443475, -0.22457938476273365, -0.035998214734935474], "ratio96": 1.985795010674685, "ratio_final": 1.9858703539332918, "diag": {"ratio": 1.9858703539332918, "nonnormal": 0.7229324901114336, "eig_gaps": [0.07164501552118119, 0.3093201053822483, 0.3519893193601399], "discness": 0.029618503748748738, "peaks": 3, "deg": 3, "diamW": 2.26041917130981}, "class": "near2-double", "probe_best": 1.9858703539332918} +{"family": "jordan2", "seed": 402005, "x": [0.5143643310319261, -0.6196694125444009, 0.4267571563656317, -0.3923463764647759, 0.6713796651131778, -0.8561063513639969, 0.9931120120064981, -0.27910508480680024, 0.0007202774379802669, 0.1049586161966505, -0.05055213341976966, -0.04996901003656072, -0.7063548096244454, 0.6677229579663577, 1.3117775801216105, 0.05423293843785007, 0.05657641294003245, 0.14307545003567024, 0.05418561276463538, -0.04348766628545159], "ratio96": 1.9966997062109073, "ratio_final": 1.996750219923216, "diag": {"ratio": 1.996750219923216, "nonnormal": 0.4824233480878822, "eig_gaps": [0.22836506835912282, 0.2660517104019978, 0.49148976810475037], "discness": 4858400643.76638, "peaks": 3, "deg": 3, "diamW": 1.0668013127831568}, "class": "near2-other", "probe_best": 1.996750219923216} +{"family": "nearnormal", "seed": 403005, "x": [-1.4703097829670948, 1.0390268841869092, -0.7865790963078486, 1.6359837168114044, 1.0877772547716655, 0.7227192705730611, -0.587643242330357, -0.895087453752738, 0.7099360984889834, -0.5174173653380031, 0.874317687508995, 0.022440906111533248, 0.687642722744501, -0.2110050131760136, -1.3888193051476074, 0.23235956995629434, 0.5934988952902791, -2.014221857227639, 1.0375516101990683, -0.018903007991207574], "ratio96": 1.3864290267504016, "ratio_final": 1.4741437723143012, "diag": {"ratio": 1.4741437723143012, "nonnormal": 0.43593527792756276, "eig_gaps": [0.29045294694821994, 0.6672076143691975, 0.8248276902568451], "discness": 1.176211502769235, "peaks": 3, "deg": 3, "diamW": 3.12497823879438}, "class": "stall", "probe_best": 1.4761142760683887} +{"family": "companion", "seed": 404005, "x": [1.296994532140058, -0.03133698333521135, -1.0254537587940642, -0.6823999212744614, 0.7978940006207635, -1.3030876603584725, 1.8818992824208098, -1.0993878647814024, 0.9364777602181457, -0.8492884361492198, 1.187205650218838, 0.8919213827350465, -0.2314021984774604, 1.125445931538735, 0.770791527432287, 0.4817233328985806, -0.1997482687141089, -0.421763205963558, 0.13744656499692748, 0.06530014923372114], "ratio96": 1.6398986289545667, "ratio_final": 1.6760626219030748, "diag": {"ratio": 1.6760626219030748, "nonnormal": 0.7645869710313646, "eig_gaps": [0.3571611103545714, 0.503540083237188, 0.6305646374399686], "discness": 0.11786194092016526, "peaks": 4, "deg": 3, "diamW": 3.8251119175538046}, "class": "stall", "probe_best": 1.8256418236017657} +{"family": "peaks", "seed": 405005, "x": [0.7468687901284015, 0.6685977493559463, 0.6814232516318232, 0.08102344001684023, 0.6994839258406955, 0.0036812118537640024, -0.24578171503023244, -0.18377825849778728, 1.4863759641204037, -0.32885719603613306, -0.12256102012716338, 0.06470633806528447, -1.4552996096992192, 1.5948410550879046, 1.0366901638462571, -2.777955189481995, -0.3264345976570303, 0.400913794879848, 0.049967038153144946, -0.15747951694503792], "ratio96": 1.9828310090158723, "ratio_final": 1.9828228939265313, "diag": {"ratio": 1.9828228939265313, "nonnormal": 0.8470431698670341, "eig_gaps": [0.046915587968391605, 0.3492297307349634, 0.3937659771855891], "discness": 7.306914098147017, "peaks": 3, "deg": 3, "diamW": 1.6928908648905165}, "class": "near2-double", "probe_best": 1.9828228939265313} +{"family": "real", "seed": 406005, "x": [-0.7281442178773666, -1.7847475897856953, -0.4594757083180713, 0.23761267847591927, -0.21532205165320678, 0.31281523499379704, 1.930518161680395, 0.35606654458631937, -0.036404096858650564, 0.4764740403307944, 2.2671069168100986, -1.0585756535121125, 0.0942644515161769, 0.9806320046735171, -0.07163695250500673, 2.0962117451617033, 0.6442501318745342, 0.7627860529481163, -0.17077023457125762, 0.08271346537950511], "ratio96": 1.9147373370903351, "ratio_final": 1.914785202186186, "diag": {"ratio": 1.914785202186186, "nonnormal": 0.6440140789038103, "eig_gaps": [0.06344635049218188, 0.5066630190824458, 0.536269467265904], "discness": 0.11937327871219945, "peaks": 3, "deg": 3, "diamW": 4.02659816689329}, "class": "stall", "probe_best": 1.9397286126115347} +{"family": "random", "seed": 400006, "x": [0.47863329172096325, 0.48793596154177526, 0.6372176039099714, 0.0284562835628297, -0.8836849554439321, -0.16841313139270264, -0.0022889081007685477, 0.3241012857486631, -0.45636832006141087, 1.2309031502173364, -0.7240709123722814, -2.009383641295884, -0.46215936889684694, 0.1655222153944077, -2.214298982073867, 0.8310458265607209, 0.10357192200359389, 0.04631930232135177, 0.07213673277293825, -0.04344062708867509], "ratio96": 1.9594436350872872, "ratio_final": 1.9603417569805628, "diag": {"ratio": 1.9603417569805628, "nonnormal": 1.2692905141439979, "eig_gaps": [0.16746401471759936, 0.5209808721524615, 0.5283556065827749], "discness": 0.05416115878220803, "peaks": 4, "deg": 3, "diamW": 2.902574031158226}, "class": "stall", "probe_best": 1.9645785577272212} +{"family": "jordan3", "seed": 401006, "x": [0.4777136052106101, 0.35248140929494287, -0.06854745245041499, 0.08456361250332894, -0.47559179222976194, -0.43222931975513346, 1.2334841037062203, -0.4902254144574247, 0.07798739444500802, 0.04137142063186865, 1.355231026190787, -0.04068449073975561, -0.061105423358001724, -0.20281475193284015, 0.0521339665120317, 0.056329094466752355, 1.1130677860472218, -0.09034012277684, 0.025158776997299598, 0.05815317951772048], "ratio96": 1.9787425436967245, "ratio_final": 1.9787114916978592, "diag": {"ratio": 1.9787114916978592, "nonnormal": 0.6969931859501752, "eig_gaps": [0.2685455033717608, 0.29035836434580736, 0.5449834719264761], "discness": 0.06219869401728481, "peaks": 4, "deg": 3, "diamW": 2.265631266900586}, "class": "near2-other", "probe_best": 1.9787114916978592} +{"family": "jordan2", "seed": 402006, "x": [-0.08890400897663925, -0.2624976354787355, -0.3779075569854964, 0.012058282264500973, -0.4037853084763352, 0.21474625495157695, 1.2670627624410689, 0.016479592820448002, -0.11605418925057426, -0.686226135665413, -0.13542610034458946, -0.046874905699384545, 0.2721266519153368, 0.11284423036045615, 1.0291075437552584, -0.1753866082358868, -0.007617948230046349, 0.028664230824509015, -0.00458893631260931, -0.030389275996942952], "ratio96": 1.99160500015283, "ratio_final": 1.9916104649078281, "diag": {"ratio": 1.9916104649078281, "nonnormal": 1.210055576623942, "eig_gaps": [0.13649143451114973, 0.26627708795018706, 0.381928253988383], "discness": 0.015489242277681945, "peaks": 3, "deg": 3, "diamW": 1.4970407416472709}, "class": "near2-double", "probe_best": 1.9916104649078281} +{"family": "nearnormal", "seed": 403006, "x": [-0.7659576633390519, 0.38975503508264864, -1.1065843864571647, 0.14879262496955314, -1.2335387214227997, -0.4383193871706421, -1.1198160147188005, -0.39220251285054286, -0.0683964730898167, -0.23225867953575263, 0.7468546987946749, 0.5397924054122228, 2.564436238746241, -5.018324259456307, 6.044052512787314, -7.186533974513433, 4.23343591797086, -1.197714964198174, 0.9565975415735437, 1.4323610757267335], "ratio96": 1.9575612428068163, "ratio_final": 1.9605124786432069, "diag": {"ratio": 1.9605124786432069, "nonnormal": 0.34992695012715297, "eig_gaps": [0.233109225988691, 0.33559639490033943, 0.5312989280678324], "discness": 46167255276.48359, "peaks": 3, "deg": 3, "diamW": 1.789891729246906}, "class": "stall", "probe_best": 1.9712469977708593} +{"family": "companion", "seed": 404006, "x": [0.6781380566474811, -0.7972147023034903, -1.572400798866932, 0.4741713429775596, -1.6618896277407154, 1.1287371524244219, 1.1810548390397682, -0.09682037845502535, 0.4569680111698647, -0.5678805894215304, 1.9246397311918149, 0.664884313606311, 0.6085245285365808, -0.25808124339802985, -0.7389591255302541, -0.3199681377431427, -0.5180025955110048, -0.592395850810844, 0.0215185141308544, -0.10230995807143764], "ratio96": 1.780125726519464, "ratio_final": 1.8022693302847397, "diag": {"ratio": 1.8022693302847397, "nonnormal": 0.5964756427411668, "eig_gaps": [0.16157436135687997, 0.6321642681671328, 0.7412043619293848], "discness": 0.4182798663194334, "peaks": 4, "deg": 3, "diamW": 4.0888585734440515}, "class": "stall", "probe_best": 1.884779979999622} +{"family": "peaks", "seed": 405006, "x": [-1.1419118386480398, -0.40955671665534976, -0.1721822826431238, -0.07778192444747811, 0.16137979685811044, 0.22121779785775608, 1.0463878612064088, 1.021243806466734, -0.2620039701378799, -0.03840032138143459, 0.12363797089089637, 0.0908202951125826, -1.4328465563240305, -2.5035457620609467, -2.4527590597342175, -1.8428283731440982, 1.5207386894886987, 0.8572447715455168, 0.6951247435177166, 0.08664003621127568], "ratio96": 1.9152487471896869, "ratio_final": 1.9212837509770775, "diag": {"ratio": 1.9212837509770775, "nonnormal": 0.9949379766720013, "eig_gaps": [0.2379285442671748, 0.5443759829055729, 0.7690473422151876], "discness": 0.4352479611365115, "peaks": 5, "deg": 3, "diamW": 1.8827329381893674}, "class": "stall", "probe_best": 1.9301790808389019} +{"family": "real", "seed": 406006, "x": [0.20028678957252943, 0.04987978972969981, 0.31328807148201576, 0.01846701689458702, 0.17800085797955956, 0.04435230193614961, 0.3191462580895724, -0.007836581848676596, -0.27672953846139803, -0.056185470385430425, -0.20248183849878898, -0.02565545274375014, -0.5740675909760251, -0.15199523572124785, 1.8934955191926033, 0.21795612371750883, 0.25283256690140277, 0.33081095676531436, 0.4651695086025137, -0.06944594864961291], "ratio96": 1.974460847202167, "ratio_final": 1.9763202426022828, "diag": {"ratio": 1.9763202426022828, "nonnormal": 0.7837545030619908, "eig_gaps": [0.04822108625623177, 0.24631431777376378, 0.2892724487887415], "discness": 1401087496.616828, "peaks": 3, "deg": 3, "diamW": 0.4761647559737326}, "class": "near2-double", "probe_best": 1.9763202426022828} +{"family": "random", "seed": 400007, "x": [-1.7271974843956723, -1.3307173372291028, 0.9342054411926182, -1.6072672721446133, -2.326967614663957, 1.7583782074919152, 2.6354734332322716, -1.282194938696977, 0.22886904940067898, -1.3560563321396204, 2.4227444424658984, 1.4369103261412157, -3.710980836557621, 13.91049036141306, 3.5479043452632757, 5.266578910299557, -0.8885590888750532, 3.1066713889548545, -2.732535811112286, -1.2251635406336698], "ratio96": 1.47935934464597, "ratio_final": 1.4825509958198984, "diag": {"ratio": 1.4825509958198984, "nonnormal": 0.7179693033103951, "eig_gaps": [0.42039853141742434, 0.49440754732533504, 0.7363130703655532], "discness": 0.1582060696059202, "peaks": 3, "deg": 3, "diamW": 6.364752691845483}, "class": "stall", "probe_best": 1.4840616347024944} +{"family": "jordan3", "seed": 401007, "x": [-0.23990780783593102, -0.45277256917452346, 0.035258983070475064, 0.40010524004005, 0.22276981934322648, 0.32671683741842095, 1.252562316244276, -0.05290650274406739, -0.42173326953585777, 0.35833500655730477, 0.9667544482577632, 0.5622353513558098, 0.0185843151238841, -0.049796182909323534, -0.12829691892727413, -0.3865614152633883, 1.2520341909045287, -0.280135177653246, 0.1110900276544754, -0.4086311989177509], "ratio96": 1.978459560993487, "ratio_final": 1.9784330459657076, "diag": {"ratio": 1.9784330459657076, "nonnormal": 0.7693817038224604, "eig_gaps": [0.09970809729324441, 0.4437567775179032, 0.44885426947901735], "discness": 0.0470015421385097, "peaks": 4, "deg": 3, "diamW": 2.0195029734736525}, "class": "near2-double", "probe_best": 1.9784330459657076} +{"family": "jordan2", "seed": 402007, "x": [0.26665838195127134, 0.0640304725161513, 0.16347612139243972, 0.12743623276967211, 0.18638742679347298, 0.146708482375494, 0.28023340580960826, 0.26874593291937754, -0.36574668595684606, -0.20622109820791154, 0.017300261346530968, 0.056087999820363825, -0.31719101950103934, -0.08482792471879236, 1.4484895367671293, -0.1559848125664955, 0.03923994409868693, 0.33784815332331863, -0.1609793261886948, -0.10474018452002712], "ratio96": 1.9921684385032334, "ratio_final": 1.9921654931823096, "diag": {"ratio": 1.9921654931823096, "nonnormal": 0.9352248673993476, "eig_gaps": [0.05126140676795281, 0.19730418455475635, 0.20735810777389252], "discness": 0.03682290617984322, "peaks": 3, "deg": 3, "diamW": 0.5840468557943083}, "class": "near2-double", "probe_best": 1.9921654931823096} +{"family": "nearnormal", "seed": 403007, "x": [-2.0958263771046246, 0.9701023708898655, 0.6164119147489941, 0.7756431918705813, 0.17083768167684252, -0.9791742802419883, -3.5641232463507273e-06, 6.772032910948186e-06, 1.1496464631278123e-05, -2.9272011800299707e-06, -0.356252371217637, -0.2690348562352274, -0.16799366373451458, 0.30846891680998456, -0.008491575676104483, -0.010693331105547498, 0.1930001992003953, -0.7225364008953097, 1.0662043126999898, 0.1764901894661851], "ratio96": 0.999999999977969, "ratio_final": 0.9999999999855319, "diag": {"ratio": 0.9999999999855319, "nonnormal": 0.1569196375646732, "eig_gaps": [0.593760806071206, 0.891771358993788, 0.9804354427669555], "discness": 0.4904988207744711, "peaks": 1, "deg": 3, "diamW": 3.049212550471817}, "class": "one-exact", "probe_best": 0.9999999999860166} +{"family": "companion", "seed": 404007, "x": [-0.07935847544128258, 0.1696381009303255, 1.1352124039824503, -0.15961559778566276, -0.39430256442774175, 0.5081402967911511, 1.8969489142449998, 0.09283326883950821, -1.1493055953530875, -0.7118155423153181, 1.1203721797830306, 1.3025041829869053, -2.3073958249794297, -0.43826862531356037, 0.36747562660831046, -2.0319113383563234, 0.968562569523401, 3.6915163677134735, 0.5861255011108146, 0.3171651183809855], "ratio96": 1.9511460663082802, "ratio_final": 1.9513804021932275, "diag": {"ratio": 1.9513804021932275, "nonnormal": 0.8012855960238126, "eig_gaps": [0.13706704183030086, 0.37305900094019234, 0.494758391595765], "discness": 0.06800897941042856, "peaks": 4, "deg": 3, "diamW": 3.373214122922498}, "class": "stall", "probe_best": 1.9626252905305392} +{"family": "peaks", "seed": 405007, "x": [-0.973207552073375, -1.4253653104832993, -0.26138240654902, 0.08474461463385796, 0.09132457192601559, 1.089651387682911, 2.314842792736478, 0.36220712431232255, 0.22948750746941293, 0.3109728179097097, 1.9483298913820557, -0.2379630867976427, -0.4562084223261708, -2.666478950685109, 1.053158648790393, -1.4215321117877782, 1.368100920932, -2.6405224726413254, -0.23838834900741335, -0.15145099907731274], "ratio96": 1.911489928763168, "ratio_final": 1.9324690744761903, "diag": {"ratio": 1.9324690744761903, "nonnormal": 0.7041259506728255, "eig_gaps": [0.25882706788689464, 0.4057286030054735, 0.6637187510134418], "discness": 0.1332300150112916, "peaks": 4, "deg": 3, "diamW": 4.114743382705791}, "class": "stall", "probe_best": 1.9516353926813574} +{"family": "real", "seed": 406007, "x": [0.4232805006186566, 0.3032863466427486, 0.1079836398073164, 0.36446421411755003, 0.2902826232154063, 0.15080706600441307, 0.05382965082075028, -0.07013063794574523, 0.2464980655309848, 0.09515916196395252, -1.0812407159761497, -1.121832427984087, -2.0454676493970085, -1.1928086162389375, 6.426534692902552, -2.9185576933211994, -0.1788615215363415, -0.05543002642855899, -0.01820183772369944, -0.07788665355238644], "ratio96": 1.9978320697756538, "ratio_final": 1.998177627046962, "diag": {"ratio": 1.998177627046962, "nonnormal": 1.1867260636607906, "eig_gaps": [0.1259377741763055, 0.17481573930682828, 0.1999104827018917], "discness": 0.0088784670732251, "peaks": 3, "deg": 3, "diamW": 1.6066054693574767}, "class": "near2-double", "probe_best": 1.998177627046962} +{"family": "random", "seed": 400008, "x": [0.3663403240379718, -1.1630356043286478, 1.2107192498068509, 0.3936918947594209, -0.038502087020308225, 0.39526120009010735, 1.1705550787976575, 1.8976072804933066, -0.06269606842011673, -1.2818891589261967, 2.0495081609776005, 1.555995408049099, 0.18351717531629252, -1.820402701816372, -1.11558274367599, 2.124843466657353, 1.3520182725911765, -1.3981797771030908, 0.28241729181316977, -0.05694894795418612], "ratio96": 1.9677583889649517, "ratio_final": 1.971935983053439, "diag": {"ratio": 1.971935983053439, "nonnormal": 0.7316288357018575, "eig_gaps": [0.3066673255202305, 0.3952399864982682, 0.4347522613966448], "discness": 0.03423094526981491, "peaks": 4, "deg": 3, "diamW": 4.073542299980285}, "class": "near2-other", "probe_best": 1.971935983053439} +{"family": "jordan3", "seed": 401008, "x": [0.27264417516194184, -0.40355358427079496, -0.7706821774908218, -0.18341435111128662, -0.23048979941639292, -0.11135955917322606, 1.2087084730597066, -0.42294884578635816, 0.17915083226326187, -0.20651481397510119, 0.9728925904740653, -1.0830945158741114, -0.24414555321492099, 0.202622436898433, 0.32658301662328304, 0.9770432965626488, 1.744250967534314, 0.7485854415286339, 0.3363813613813472, 0.029639818913553655], "ratio96": 1.9849440399298166, "ratio_final": 1.9851595972644236, "diag": {"ratio": 1.9851595972644236, "nonnormal": 0.6776964068542172, "eig_gaps": [0.2453687395485127, 0.2619594851466761, 0.480086807418659], "discness": 0.06793844681529179, "peaks": 4, "deg": 3, "diamW": 2.221052233486238}, "class": "near2-other", "probe_best": 1.9851595972644236} +{"family": "jordan2", "seed": 402008, "x": [0.3519641992289182, -0.04579720347873614, -0.09893467737384185, -0.008689455604605364, 0.4718924716103322, -0.08844095784534273, 0.8307727729276292, -0.24756938141068757, 0.25568642097485117, 0.2574770640178582, -0.19787738215038045, -0.2543383852506127, -0.17870936527071085, -0.0570355804312576, 1.349151372773137, 0.7681295252837211, 0.0756724444141999, 0.028864119031639504, -0.18885460152250347, -0.1428852586742786], "ratio96": 1.9832085194658557, "ratio_final": 1.9832604521619137, "diag": {"ratio": 1.9832604521619137, "nonnormal": 1.1150821174767303, "eig_gaps": [0.11798036972872193, 0.41935322170949074, 0.5342413163740265], "discness": 0.038059459484859864, "peaks": 4, "deg": 3, "diamW": 1.0788595620742354}, "class": "near2-double", "probe_best": 1.9832604521619137} +{"family": "nearnormal", "seed": 403008, "x": [-1.2136214967580576, -0.5142285823882673, -0.1465749602247652, -0.06667532011424951, -1.4708485482677087, -0.6872232275816879, 0.008563858765450022, -0.004461839829838391, -0.5851578828624939, -0.3969130031179755, -0.011592313923957662, 0.004905244349369347, 1.0739739406249045, 0.8834791741412173, 0.29789187133457556, 1.8413012504767763, 1.8196464476688146, 1.050587273626323, 0.9967483982229499, -0.44381995529477514], "ratio96": 1.7180459040155245, "ratio_final": 1.6624134520100349, "diag": {"ratio": 1.6624134520100349, "nonnormal": 0.1577304130451886, "eig_gaps": [0.1829134951908633, 0.6827675971972664, 0.8629451110530018], "discness": 189183537176.33398, "peaks": 1, "deg": 3, "diamW": 1.6947276261115944}, "class": "stall", "probe_best": 1.7489506964823196} +{"family": "companion", "seed": 404008, "x": [2.664009232228862, -0.3456109942432075, 4.972441164947222, 0.14076840622208842, -0.962158746555563, -0.024445359725746282, 5.072146651714034, 0.1659179765008987, -3.0830970186520457, -1.4347715708696072, 4.446036622431034, 2.825890644636497, -1.5096694781141298, 0.5443261783400037, 1.4797321859270802, -3.755333038337281, -0.3871979566872791, 0.6381976898419595, 0.035510502723660964, 0.046969482821620556], "ratio96": 1.899294501088215, "ratio_final": 1.9150984233362012, "diag": {"ratio": 1.9150984233362012, "nonnormal": 0.6237362157721753, "eig_gaps": [0.2364953070268628, 0.3649371813749639, 0.595159149041349], "discness": 0.127041889561606, "peaks": 4, "deg": 3, "diamW": 9.97531361954089}, "class": "stall", "probe_best": 1.9172231986490849} +{"family": "peaks", "seed": 405008, "x": [-1.1314193435648412, -1.00250572596297, 0.11063819796007958, 1.2577304216756229, 1.27383486195379, -0.3335212297564938, 2.2979015244603787, -0.39986638719297934, 0.12205146689797608, 0.39629482333413946, 2.020309128553821, -1.2887045489219133, 2.995757357409451, 0.3250184146572523, -0.8128331833086984, -1.1728612113036694, -2.5749123866084087, 1.7710169154296285, -0.37568249439298873, -0.48005946119030796], "ratio96": 1.9000558645184835, "ratio_final": 1.9005921664070224, "diag": {"ratio": 1.9005921664070224, "nonnormal": 0.7383390533156136, "eig_gaps": [0.4562802022192868, 0.5779246784507605, 0.5970157071751007], "discness": 0.050636755992918676, "peaks": 4, "deg": 3, "diamW": 4.31986265628844}, "class": "stall", "probe_best": 1.9291540674642105} +{"family": "real", "seed": 406008, "x": [0.659656126983944, 0.15189431804171882, -0.4703473851413998, 0.15994728395246607, -0.5600274361703954, -0.8525039622187867, 1.3899332166086935, -1.6762096475073651, 1.0697186474376594, -0.10308842923344212, -0.2797099957286665, 0.6333234098019448, -0.978903263955067, -1.1646340066165992, 1.9510508922728858, 1.2181426055373, 0.5070068044118878, 0.365328149243664, 0.12117339301094164, 0.07734093893904068], "ratio96": 1.7606056140911168, "ratio_final": 1.7689261204340256, "diag": {"ratio": 1.7689261204340256, "nonnormal": 1.1700312175947936, "eig_gaps": [0.36620284600231395, 0.4071377270911714, 0.56926100085038], "discness": 0.03279861213723839, "peaks": 3, "deg": 3, "diamW": 2.775552672435289}, "class": "stall", "probe_best": 1.7697799595903716} +{"family": "random", "seed": 400009, "x": [-0.22774092893790934, -0.1216641819292445, -0.4204469161047361, 0.7866393496535915, 0.288704211557278, -0.45537575151179077, 0.05322850940166426, -3.4211980300379783, 0.36298019870612186, 0.40746809888878244, 2.0735396358383764, 2.5748028511991663, 0.2701768202932438, -0.2893503324187542, -0.1646704355750226, -0.32230525695657564, -0.045098664957112185, -1.765473364603786, 0.03362495212871314, -0.05559961284806482], "ratio96": 1.9946504192233194, "ratio_final": 1.9946369355893576, "diag": {"ratio": 1.9946369355893576, "nonnormal": 0.7129899058436834, "eig_gaps": [0.12305906438616224, 0.1858292087563823, 0.28623434277059073], "discness": 0.015763486225498797, "peaks": 3, "deg": 3, "diamW": 4.996635462866935}, "class": "near2-double", "probe_best": 1.9946369355893576} +{"family": "jordan3", "seed": 401009, "x": [0.03854468427709147, -0.21022928805580704, 0.0506578546619329, 0.38943210640504344, 0.09142038511193193, -0.13194916233997026, 1.2970898211924482, -0.18152656507851092, -0.0215318444762966, 0.15296657651060486, 1.2721598273369157, 0.407468069685303, 0.04422913860446514, 0.008915778283282806, -0.13939731143541584, 0.024819265834887878, 1.366880516446983, 0.04509611587755551, -0.041354312861967266, 0.22256727134678608], "ratio96": 1.9947657274448514, "ratio_final": 1.9947499524268297, "diag": {"ratio": 1.9947499524268297, "nonnormal": 0.728721986424458, "eig_gaps": [0.04775604062686137, 0.26438481408176895, 0.3032162756641244], "discness": 0.020291685796728715, "peaks": 3, "deg": 3, "diamW": 1.978072330693802}, "class": "near2-double", "probe_best": 1.9947499524268297} +{"family": "jordan2", "seed": 402009, "x": [0.22183406818414309, -0.3321832194025729, -0.40849962038569954, -0.06922118732437961, 0.278497682336157, 0.09980074439975051, 1.2826594794200945, -0.10596219242046855, -1.6563503834895157, 0.42080913091138317, 0.9945353503740935, 0.050362417825558975, 0.026560965590623903, 0.3714469160903635, 2.040100655174465, 0.8938034619853016, 0.042665425660646364, -0.012651063073998527, -0.007046327805114161, -0.017651152478195452], "ratio96": 1.9902312035846754, "ratio_final": 1.9904085835769856, "diag": {"ratio": 1.9904085835769856, "nonnormal": 1.3661241825453168, "eig_gaps": [0.17799973732628177, 0.27903520714674857, 0.2890440277791622], "discness": 0.011276716001756507, "peaks": 3, "deg": 3, "diamW": 2.4476688691882047}, "class": "near2-other", "probe_best": 1.9904085835769856} +{"family": "nearnormal", "seed": 403009, "x": [-2.261226861468677, -0.3003322336238823, -0.022968297858582024, -0.420352629746012, 1.5824849181249516, 0.3742987391553157, 0.8796882768131438, -1.7375613588103362, 1.4582913741048849, -0.6289426982009182, -1.9070421162403446, 1.5427567451418556, -2.738006525376435, 4.990450730579992, 0.06665217112313647, -1.6327440351293863, 1.1137290167196339, -2.4593881059189684, 0.36750743242940986, 0.13791377050920234], "ratio96": 1.7749212299156645, "ratio_final": 1.7786737363778258, "diag": {"ratio": 1.7786737363778258, "nonnormal": 0.836273567732578, "eig_gaps": [0.3408617009768091, 0.4265111679550962, 0.7425674253152206], "discness": 0.17031955619375466, "peaks": 3, "deg": 3, "diamW": 5.255370321655844}, "class": "stall", "probe_best": 1.7833842649461549} +{"family": "companion", "seed": 404009, "x": [-0.47953131525842196, 1.0135699649509924, 1.1274456053034565, 0.8229293234116888, 0.447596741949128, -0.9545088697078681, 1.7048672075918638, -0.05689587655826359, -1.4568181404402187, 0.5310104002156508, 1.6189119590903311, 0.5436015078068784, -0.9395735625339497, 0.8987905473702422, 0.1402145709363042, -1.9167010744421695, 0.7546786102828538, 1.8287537229948052, 0.2505592572055595, 0.20412091873636218], "ratio96": 1.7923898947444397, "ratio_final": 1.795200320138665, "diag": {"ratio": 1.795200320138665, "nonnormal": 0.7703200637539882, "eig_gaps": [0.448773949957648, 0.5277476674972144, 0.6033188776751638], "discness": 0.05210113169966838, "peaks": 3, "deg": 3, "diamW": 3.6059257795764674}, "class": "stall", "probe_best": 1.8623713093222618} +{"family": "peaks", "seed": 405009, "x": [-0.7068786440132873, -0.09410910025856656, -0.08561227921145037, 0.37625518561316074, -0.1519356983949856, -1.091822262573443, 2.3215219516527306, -0.48160579471035286, 0.34479439496156417, 0.4226598359761319, 2.105438267215958, -0.6627049716399123, 0.5387849201432447, 2.0015857003613235, 1.495419116298284, 3.9930089678493683, 3.562255546932391, 1.8349674222382366, -0.47608679349979105, -0.035811581855137074], "ratio96": 1.9658277206488137, "ratio_final": 1.9657348708525542, "diag": {"ratio": 1.9657348708525542, "nonnormal": 0.6919349364269667, "eig_gaps": [0.21575333661915772, 0.3160997840111913, 0.4068910854583067], "discness": 0.03404861946510586, "peaks": 3, "deg": 3, "diamW": 3.6117154824153883}, "class": "stall", "probe_best": 1.9707873277970684} +{"family": "real", "seed": 406009, "x": [-0.37284670761904504, 0.0293887097336996, -0.018522783882466948, 0.006435549263959043, -0.3373520703591677, 0.004326600233458357, 0.4147470295667798, 0.03716035756059122, 1.3144811487381274, -0.010562581081042847, 1.3253691352061527, 0.054628003139014475, 0.4580649006121915, -0.08106992302865976, 2.8900650306552955, 0.03209881855142008, 0.009313055527221059, -0.016734127394161094, -0.01619871494432045, -0.007460463057680774], "ratio96": 1.998642770727212, "ratio_final": 1.9986423314570214, "diag": {"ratio": 1.9986423314570214, "nonnormal": 1.3348378307665354, "eig_gaps": [0.022462406332965982, 0.1648258988376734, 0.18355556940933307], "discness": 0.004919149520888344, "peaks": 3, "deg": 3, "diamW": 1.934382057946099}, "class": "near2-double", "probe_best": 1.9986423314570214} +{"family": "random", "seed": 400010, "x": [4.856639551490209, 3.6646571690516847, 2.8745685127335445, -1.9338118723509474, -1.6671272273111688, 3.694518307049382, 1.4208738587602596, -4.658863316165691, -1.6523013114433969, -3.229457436235249, -3.746301993855038, 6.876943177885096, 11.390685949573218, 5.156109842632173, -10.657682300350091, -8.20424004032496, 0.8686878220323329, -1.747136165628734, 0.7089508389620098, -0.2525604106082646], "ratio96": 1.4880523186988779, "ratio_final": 1.4882500726649115, "diag": {"ratio": 1.4882500726649115, "nonnormal": 0.7863933697500112, "eig_gaps": [0.4858696563107541, 0.5337169895193274, 0.5916707133714624], "discness": 0.0556075991695392, "peaks": 3, "deg": 3, "diamW": 12.223397883181907}, "class": "stall", "probe_best": 1.4891708302973052} +{"family": "jordan3", "seed": 401010, "x": [-0.3149361587408003, -0.21551196733305505, 0.339050491458098, 0.6056176203228842, 0.07777994851851322, -0.44507405430720837, 1.1161723718165852, -0.5003681068699157, -0.09718400664794431, 0.8295971120324027, 1.3301641032656117, 0.32579238355274726, -0.09525119994046503, -0.21063944441270524, -0.08260118119993072, -0.05262079689801061, 0.8464161957684946, -0.9464314057087773, -0.2535401813273302, -0.055805525021297694], "ratio96": 1.9638052265768613, "ratio_final": 1.9638134664520799, "diag": {"ratio": 1.9638134664520799, "nonnormal": 0.7285249786253036, "eig_gaps": [0.19389124821095202, 0.4474386051853289, 0.4614830330634066], "discness": 0.05053047081210302, "peaks": 4, "deg": 3, "diamW": 2.3461077434622393}, "class": "stall", "probe_best": 1.9767128808328067} +{"family": "jordan2", "seed": 402010, "x": [-0.4705287650038279, 0.42166434140105413, -0.1718318218675568, 0.4510703707222654, 0.49956016229896766, 0.4543597439405777, 0.8746949028950749, 1.0023744599041588e-05, -0.0035285305848654328, 0.004626425329218299, 0.005019851121867839, 0.0008216205316423008, 0.19711407940523973, -0.6161875452120507, 0.997986660860545, -0.06507005689542798, -0.23505349471640766, 0.5341524006395462, -0.1820064684907137, 0.3809988618241933], "ratio96": 1.872399103464923, "ratio_final": 1.8562662258619247, "diag": {"ratio": 1.8562662258619247, "nonnormal": 0.6172643776303153, "eig_gaps": [0.2338984695993838, 0.5232190152823144, 0.7564151627318185], "discness": 1.8301968914279765, "peaks": 4, "deg": 3, "diamW": 1.283210323738045}, "class": "stall", "probe_best": 1.880302827415056} +{"family": "nearnormal", "seed": 403010, "x": [-0.9610754374302735, -0.6719738245615725, -0.7857233904447865, -0.5497906539944826, -1.1400208942459042, -0.8957377802591684, -0.02442360831907832, -0.0767926443835779, 0.7139188044994125, 0.00374520667182384, 0.05566666916645961, -0.12969757516720723, -1.7510823291064395, 0.4279630895602277, 1.0259909544685493, 3.0928670423848805, 2.69394800443116, 1.8801664219633998, 1.093431860446832, 0.49782049463338446], "ratio96": 1.8408427699229664, "ratio_final": 1.8421403904378362, "diag": {"ratio": 1.8421403904378362, "nonnormal": 0.16338197271246674, "eig_gaps": [0.2687331934158094, 0.36026498458146017, 0.6226412251172806], "discness": 11422575498.476671, "peaks": 3, "deg": 3, "diamW": 0.7952941617792536}, "class": "stall", "probe_best": 1.8641011856220107} +{"family": "companion", "seed": 404010, "x": [-0.23358967559571897, -1.28090873738949, -0.7468567890202502, 1.3670106960780437, -0.20804909800634963, 1.9811735913249477, 2.4923266400198605, -0.7965237822350789, 0.001541559998433733, 0.48725428175651775, 2.6326447064789273, -0.11277816185833346, -1.4768915993350487, 1.1615159123621241, -0.06448308105005324, 2.080310714376713, -1.261041107460945, 0.8584670953106981, -0.08914025674475612, 0.05629974699860085], "ratio96": 1.92877613632646, "ratio_final": 1.9287354498093967, "diag": {"ratio": 1.9287354498093967, "nonnormal": 0.6416282133399552, "eig_gaps": [0.16037269234467313, 0.5294387506893278, 0.6403388167155266], "discness": 0.1629769584946493, "peaks": 4, "deg": 3, "diamW": 5.0944628490555015}, "class": "stall", "probe_best": 1.9296156605938781} +{"family": "peaks", "seed": 405010, "x": [0.24024679289498257, -0.09731390730206357, 0.03210074607578857, 0.8053772392538259, 0.8965040750705746, -0.09321264765640486, 1.2196734180511108, -0.44142246410435737, -1.1773530493057107, -1.8214142335284798, 2.04247424093146, 0.8308718314546747, 0.1968334565743245, 1.5307461436804362, -4.0126503716519135, -1.0851022404481183, 0.2776240211350127, -0.004356946561595831, -0.016512687924860693, 0.02821681622848994], "ratio96": 1.9707837571591873, "ratio_final": 1.9707916175806544, "diag": {"ratio": 1.9707916175806544, "nonnormal": 1.3046358344313416, "eig_gaps": [0.18618126605255556, 0.26280980637326545, 0.35372898041396844], "discness": 0.019388377633169262, "peaks": 3, "deg": 3, "diamW": 3.524898671668976}, "class": "near2-other", "probe_best": 1.9707916175806544} +{"family": "real", "seed": 406010, "x": [-0.13682990206278445, 0.3913885021690906, -0.12074141731529525, -0.07047287059811458, 0.13528976648032126, -0.02216296556349686, 0.188846236127049, 0.06948991252943004, -0.16271124501777967, 0.4279605114350058, 1.206258133411756, 0.2073399997472349, 0.2083695558312193, -0.15661107949724834, -2.9246889293417375, 0.48414049450203656, -0.4418169137246779, 0.1189526676206738, -0.017677713428430462, -0.00890760876699348], "ratio96": 1.9787685151411023, "ratio_final": 1.9787659977707426, "diag": {"ratio": 1.9787659977707426, "nonnormal": 1.2797728668030108, "eig_gaps": [0.19336734652642415, 0.34297983996117887, 0.3674025622666667], "discness": 0.0068932916331201095, "peaks": 4, "deg": 3, "diamW": 1.347430511092709}, "class": "near2-other", "probe_best": 1.9787659977707426} +{"family": "random", "seed": 400011, "x": [6.684187815291834, -8.914533312051766, -7.443576013534653, 2.361941844938748, 7.354335432749464, 9.295295938199336, -4.33093454663995, 30.381515223324296, -4.576483796722105, -5.200174912813532, -34.38577199236677, 1.0975426904637107, 1.4566245048452813, -5.063322773548105, -20.10264502690636, -6.5022029921920055, 53.10871810533355, -3.2517042157857032, -0.10758533057022972, 0.3926349548519319], "ratio96": 1.9019796032179987, "ratio_final": 1.9121754407897367, "diag": {"ratio": 1.9121754407897367, "nonnormal": 0.7552378044154928, "eig_gaps": [0.32786444476236953, 0.36266688210669373, 0.36559322379457593], "discness": 0.008152510539096495, "peaks": 3, "deg": 3, "diamW": 49.842707918740025}, "class": "stall", "probe_best": 1.9304331500817857} +{"family": "jordan3", "seed": 401011, "x": [0.21351911923788608, 0.1506909270714713, -0.1768068763991678, -0.26586436515096806, 0.19310643202758504, 0.41838348010739074, 0.9157000619058213, -1.007008298246707, -0.3218669803707857, 1.1988710685927915, 0.5408501367932514, 0.4585008897582016, 0.10595034392119682, -0.07458772262724345, -1.4050802608772703, -0.7432569106861454, 0.017215053345738224, -0.034941175086837584, 0.014330583854682551, -0.03521719726700879], "ratio96": 1.9921677503216075, "ratio_final": 1.9921639591736189, "diag": {"ratio": 1.9921639591736189, "nonnormal": 1.3441349897745958, "eig_gaps": [0.12964834997104202, 0.2756739310870979, 0.37563021273762703], "discness": 0.01813847966604149, "peaks": 4, "deg": 3, "diamW": 2.0707529389739596}, "class": "near2-double", "probe_best": 1.9921639591736189} +{"family": "random", "seed": 500000, "x": [2.2010809995628335, -1.9464386842726633, -1.2698301666292287, -0.89428803019729, 1.3610875358971162, 1.7102617018999196, -2.8252671005099472, 2.5412514244086974, 1.8303891128175163, 0.9336926885547425, 0.9446971997773101, -3.7558848154159383, 0.20061937057321858, -0.48227195365050324, -1.6462906776973192, -1.0902350573899942, 0.31713912319027393, 1.0154481475251607, -0.0777387395391915, -0.018374794815367405], "ratio96": 1.8253779590675354, "ratio_final": 1.8845616755755084, "diag": {"ratio": 1.8845616755755084, "nonnormal": 0.689697639320076, "eig_gaps": [0.5258708856485272, 0.5367748155103692, 0.5440038018508758], "discness": 0.0283601437991864, "peaks": 3, "deg": 3, "diamW": 6.896898606843152}, "class": "stall", "probe_best": 1.9522089892686216} +{"family": "jordan3", "seed": 501000, "x": [-0.09798776276255913, -0.17362165051916933, 0.40148514812217984, 0.0713509880183586, -0.26531901460044605, -0.22642863415543446, 1.4657720102721816, 0.26626103947855356, 1.152503973526167, -0.10149419229378527, 0.8898035635065351, 0.012078541125608914, -0.27692657742021753, 0.17413333191125924, 1.514521996616328, -0.7713745301452021, -0.06509784928808299, -0.03800720597360517, 0.011888901907936865, 0.007039782768977668], "ratio96": 1.9786970011397063, "ratio_final": 1.979742431611181, "diag": {"ratio": 1.979742431611181, "nonnormal": 1.3073651285101924, "eig_gaps": [0.08293634463974697, 0.26294898946821815, 0.3451738207769734], "discness": 0.00971227306109078, "peaks": 3, "deg": 3, "diamW": 2.115671050451388}, "class": "near2-double", "probe_best": 1.979742431611181} +{"family": "jordan2", "seed": 502000, "x": [-0.07297242819679475, -1.1017837014494511, -0.007237127647621706, -0.1556690019805843, -0.39838040389296564, 0.2225268261918069, 1.749410684025755, 0.2920917566448649, -0.06084562506236932, -0.18798559834238168, -0.10934705833463322, -0.053172035158510794, 0.2537801433106527, 0.8864028531664594, 1.259477795852082, -0.2968643529849976, 0.0454469791824586, 0.23305786443657373, 0.10118835091616929, -0.007421163893165838], "ratio96": 1.9722743730220806, "ratio_final": 1.9725207646608844, "diag": {"ratio": 1.9725207646608844, "nonnormal": 1.096524119175018, "eig_gaps": [0.2679904453911002, 0.4671372758308442, 0.6716996727205548], "discness": 0.13265552335264527, "peaks": 4, "deg": 3, "diamW": 2.030228837510328}, "class": "near2-other", "probe_best": 1.9725207646608844} +{"family": "nearnormal", "seed": 503000, "x": [0.9514264593789671, -2.347125623512057, -0.8134464636237098, -0.8969761677742097, 0.6396030450580132, 0.541094763399647, -2.6685891043895647e-07, -2.722830965954275e-06, 1.3735402977143419e-06, -1.1667995633029469e-05, 0.13387084833551727, 0.0482949758842278, -0.02068051617361558, -0.2824950875162474, -1.0302531355781037, -1.187731093383067, -0.9359994321335581, 1.700695992494631, 1.1830909467027557, 0.10284111740736392], "ratio96": 0.9999999999811986, "ratio_final": 0.9999999999897815, "diag": {"ratio": 0.9999999999897815, "nonnormal": 0.0479452208509749, "eig_gaps": [0.7027770289696583, 0.7852365503923593, 0.9986362912686834], "discness": 1.6661929448049169, "peaks": 1, "deg": 3, "diamW": 2.9089714381264984}, "class": "one-exact", "probe_best": 0.999999999990075} +{"family": "companion", "seed": 504000, "x": [0.9775374278349032, 1.233439622907949, 0.46796876025924433, -0.37387983350426407, 0.8599935382002286, 1.0419597960369855, 0.0006866978905461961, 0.0007649866345611985, 0.12266955355014929, -0.6992793978376874, -0.0003775349167190945, -0.0008362368203637133, 0.7776427822991061, -0.28397222075708883, 0.7361749974675182, -0.7845477438405442, -0.7477518662234235, 0.8274827940607006, 0.4850179878508275, -0.4734185788720944], "ratio96": 1.7900755250285851, "ratio_final": 1.7905543163271889, "diag": {"ratio": 1.7905543163271889, "nonnormal": 0.1447556930687938, "eig_gaps": [0.11530365285323353, 0.7539336519151942, 0.8653215624718797], "discness": 230979510196.4383, "peaks": 3, "deg": 3, "diamW": 1.9485937092487953}, "class": "stall", "probe_best": 1.8040767674465368} +{"family": "peaks", "seed": 505000, "x": [-1.0606376079384545, 0.45599240756357196, 1.1951877362739876, -0.18073502213640832, 1.1539411157667614, -0.16273511406503516, 1.3692730439940373, 0.8384668224592727, -0.09011065039352059, -0.6176611892487054, 1.5564899012580486, 0.40480261342323703, -1.0652268987585274, 0.9051772403391852, -1.3177682708245633, 0.1621699844562733, 1.7433215618307192, -0.4991079651114599, 0.08341153388957076, 0.16485957998273681], "ratio96": 1.8785919580536365, "ratio_final": 1.8785472999010968, "diag": {"ratio": 1.8785472999010968, "nonnormal": 0.6487491843464325, "eig_gaps": [0.012723597949459922, 0.6500990267993801, 0.6627021766824078], "discness": 0.18383000428868004, "peaks": 4, "deg": 3, "diamW": 3.5369803068350807}, "class": "stall", "probe_best": 1.9303009251816947} +{"family": "real", "seed": 506000, "x": [1.965312424072731, 0.14450921856776555, -0.7837662964051677, 2.158561598058486, -1.6873611609670132, -2.6394955686864647, 1.7115396318888432, 3.3705297582842606, 2.760559106144206, -0.5709724700319296, -4.335412709790507, -2.312499203761174, -2.2406932866966707, -0.03694865485921059, 2.0871018989102295, -1.6745822718972447, 2.684742672867243, 4.716330459644395, 0.4848585542160312, 0.002513386941721851], "ratio96": 1.893569356397994, "ratio_final": 1.8950095242900347, "diag": {"ratio": 1.8950095242900347, "nonnormal": 0.7762494495310546, "eig_gaps": [0.4074353780503413, 0.5490816903607489, 0.5837195223607875], "discness": 0.04476857709236547, "peaks": 4, "deg": 3, "diamW": 8.364292229525603}, "class": "stall", "probe_best": 1.9286235690331417} +{"family": "random", "seed": 500001, "x": [-0.32451703623452827, -0.018664267400409797, 0.277264828121787, 0.44945329734371164, 0.05153146962232798, 0.09355566721037682, -1.4437452797936519, 0.3695577566198392, -1.4041161410296854, -0.06531523128994027, 0.035661655312237195, 0.5374024988486452, -0.7035202236342175, -0.40332272610665265, 1.427761317832099, -2.6445800620558955, 0.03772672940714584, -0.08277529345984677, 0.054159864026256894, 0.062406033866259605], "ratio96": 1.9896281567697067, "ratio_final": 1.989647419435718, "diag": {"ratio": 1.989647419435718, "nonnormal": 1.3661119810966214, "eig_gaps": [0.17452549962143868, 0.1874280866260705, 0.33906374830300945], "discness": 0.03013954368771768, "peaks": 4, "deg": 3, "diamW": 2.2485867021911887}, "class": "near2-other", "probe_best": 1.989647419435718} +{"family": "jordan3", "seed": 501001, "x": [-0.06676528698358883, -0.03143590796544164, -0.1862869641929027, -0.15728533665334532, 0.10536666765344921, 0.34613509060450565, 1.252945153974515, -0.18138884355036575, 0.13118638229781088, -0.1771079345626435, 1.163287401364013, -0.3058706140517689, 0.012256852545627674, -0.07440279612666323, 0.10273556270880962, -0.192113039623867, 1.3663520279657484, -0.3874121295125126, -0.09032172719878143, -0.06681052355270811], "ratio96": 1.99531778862005, "ratio_final": 1.9953155484498082, "diag": {"ratio": 1.9953155484498082, "nonnormal": 0.7101638555128271, "eig_gaps": [0.09337658316278655, 0.223248336538452, 0.31301176206977355], "discness": 0.020631702259158665, "peaks": 3, "deg": 3, "diamW": 1.8587235834505667}, "class": "near2-double", "probe_best": 1.9953155484498082} +{"family": "jordan2", "seed": 502001, "x": [0.7907785478031975, -0.9009602761318234, 0.5102374267968905, -0.43668580757044073, 0.182582783663714, -0.6664152161258943, 1.2315521710918673, 0.48131274895022175, -0.7395842030181672, 0.05545628501429549, -0.2922977563642658, 0.042636907089796894, -1.030466776080794, 0.7486095714866186, 1.1409348229897498, 0.10043195825595105, 0.0015796338564985565, 0.25865534248810496, 0.04909794121084454, -0.04148089289606957], "ratio96": 1.9835916006758831, "ratio_final": 1.9835924689440168, "diag": {"ratio": 1.9835924689440168, "nonnormal": 0.7376664048897879, "eig_gaps": [0.24513483313107046, 0.33229640730883725, 0.3993140885011319], "discness": 17384863855.749035, "peaks": 4, "deg": 3, "diamW": 1.6324339473517828}, "class": "near2-other", "probe_best": 1.9835924689440168} +{"family": "nearnormal", "seed": 503001, "x": [0.16209843070403318, 0.45885606673109447, -0.7183997323539497, 1.5498397045902137, -0.003656868732838745, 0.1712607253475316, 0.2799845426959468, 0.28001111064504497, -0.3160310201274161, 0.33523921268425344, -0.21999066712422746, 0.2600019699593843, -0.6478125224995749, -0.44869525649195846, -0.9379058938590823, -0.712699571399069, 0.23269225077849276, -1.402819016038616, 0.5608558604228999, 0.16911728767135858], "ratio96": 0.9999999999696355, "ratio_final": 1.2480859785227605, "diag": {"ratio": 1.2480859785227605, "nonnormal": 0.3552450807007695, "eig_gaps": [0.18678369980861795, 0.7888872768592726, 0.873785735023703], "discness": 184295262488.01663, "peaks": 2, "deg": 3, "diamW": 1.7771498014875504}, "class": "stall", "probe_best": 1.2588683347584018} +{"family": "companion", "seed": 504001, "x": [1.3454452487919029, 1.1922288230586457, -0.9711130930480547, -3.194161177589338, -0.37723292929322705, 2.61869869868932, 3.2145255716869676, -5.0837138278546465, -1.1167841449543592, -1.8837499413215197, 5.906706474268496, 0.40365313533084324, -1.419202166236122, 1.2050451450162998, 2.3904933750071438, 1.3821567296856507, -1.038057552541133, 1.0553134327493976, 0.05027133809735837, 0.09842091228662767], "ratio96": 1.9198559358004614, "ratio_final": 1.9198552462314853, "diag": {"ratio": 1.9198552462314853, "nonnormal": 0.7179767188776328, "eig_gaps": [0.2102421814648462, 0.466291139510505, 0.5492546702574377], "discness": 0.06235399989261289, "peaks": 4, "deg": 3, "diamW": 10.638268180538134}, "class": "stall", "probe_best": 1.953899344470567} +{"family": "peaks", "seed": 505001, "x": [0.7162312206223256, -0.5678662675564179, 0.05469255179749438, 0.40709325674373453, 0.8472446263569182, 0.44327986071337255, 2.2825714050836785, -1.2047064402323389, 0.16133187207321162, 0.2166060771203838, 2.554176143583713, 0.7620838388542446, -3.4356218662527964, -0.4414995512905575, 7.503203744328344, 0.38559047016323145, -5.561526836737574, 0.4092303040255224, -0.5888507723656181, -0.21526820440409136], "ratio96": 1.9920052751918345, "ratio_final": 1.9919948550362034, "diag": {"ratio": 1.9919948550362034, "nonnormal": 0.6826737683836362, "eig_gaps": [0.2029986789589923, 0.2608809458804492, 0.3014643791181786], "discness": 0.014751377828202242, "peaks": 4, "deg": 3, "diamW": 3.908290233612499}, "class": "near2-other", "probe_best": 1.9919948550362034} +{"family": "real", "seed": 506001, "x": [-0.2565006746329338, -0.2499159796967345, 0.18278277154186506, 0.8909611830041402, 0.22778763762963034, 0.29424936447882677, -0.011571650738213852, 0.16146729510346097, 0.7800526918537881, 0.05282602667020613, 3.8830013400074908, -0.975881773003448, -0.651445956736074, -0.9809025663521542, 1.863644370374139, -0.490594699480808, -0.0028718027702500978, -0.001071210878220749, 0.0024934730725696295, -0.0005913766766596903], "ratio96": 1.9986249594658858, "ratio_final": 1.9986293893348943, "diag": {"ratio": 1.9986293893348943, "nonnormal": 1.3428985634832233, "eig_gaps": [0.14481573915829782, 0.17628848170234968, 0.29585412629343966], "discness": 0.005808180815344595, "peaks": 3, "deg": 3, "diamW": 4.13219295958844}, "class": "near2-double", "probe_best": 1.9986293893348943} +{"family": "random", "seed": 500002, "x": [1.079166601601878, -0.5947810552038502, -0.01456172020084387, 0.6732708502772535, 0.05408658682694999, 0.29683281431082253, -1.9216586489470302, -1.2065669928043672, 0.4569821742136546, 0.18507263046558142, 0.8770564186705423, 2.619828477181286, -0.9310665059485426, 0.09905141295637185, 0.8535636164437238, -0.938871811718955, -0.49065554969071146, 1.2092072668239138, -0.04494175098720074, 0.15162035993734113], "ratio96": 1.9823948360410084, "ratio_final": 1.9825954160573804, "diag": {"ratio": 1.9825954160573804, "nonnormal": 0.7184301990431327, "eig_gaps": [0.0954267889141793, 0.3388136705109138, 0.41761595480531694], "discness": 0.044991606393835114, "peaks": 4, "deg": 3, "diamW": 4.009841435655586}, "class": "near2-double", "probe_best": 1.9825954160573804} +{"family": "jordan3", "seed": 501002, "x": [0.14126090417176138, -0.2257875737725679, 0.15072544542741897, 0.4887234525794647, -0.21582611246277267, -0.7768366437564411, 1.2178109970394457, 0.26884267596875044, -0.5380223155210195, 0.7968889033477157, 1.2136607390013092, 0.021033706287907485, 0.15643658027490032, -0.22369292829645748, 0.06364145519185843, 0.46519547776621406, 1.7441812814206237, -0.7985980083306673, 0.009155360496421808, -0.3706333013958898], "ratio96": 1.9746908601982593, "ratio_final": 1.9746841950990968, "diag": {"ratio": 1.9746841950990968, "nonnormal": 0.742903002605996, "eig_gaps": [0.27759081101808597, 0.3020854144965906, 0.5570034497313459], "discness": 0.078523427170913, "peaks": 4, "deg": 3, "diamW": 2.3654690819787434}, "class": "near2-other", "probe_best": 1.9746841950990968} +{"family": "jordan2", "seed": 502002, "x": [-0.5624697530949847, 0.4541501469672754, -0.3195251898566118, 0.3606021720878767, 0.9847758248860997, 0.15617389101677298, 1.2613524298033836, -0.9400003824672798, 0.019316578113084563, -0.11086746512944498, 0.02004632513240958, 0.09036710814475905, 0.17690553075938986, -1.0447647558288553, 1.1940054661990773, -0.6366325070160492, -0.33990221657859965, 0.35820916522456925, -0.11023382899490108, 0.12635023147027286], "ratio96": 1.8504650973084025, "ratio_final": 1.8437007000903027, "diag": {"ratio": 1.8437007000903027, "nonnormal": 0.8428970532761042, "eig_gaps": [0.11571003668080475, 0.5867991405895749, 0.7003401477798117], "discness": 0.2876617934303926, "peaks": 4, "deg": 3, "diamW": 2.2498741979398282}, "class": "stall", "probe_best": 1.853758779156412} +{"family": "nearnormal", "seed": 503002, "x": [-0.9905575749733214, 0.7146753300710389, 0.5608119001140869, 1.2287606375297375, -0.11492334522720735, -0.7164634425604411, -0.22272187601176902, -0.058908899535744054, -0.5615770039077239, 2.06185892988116, 0.0670065399924976, 0.06634000700909656, -1.0620917050512833, -2.3400728879323456, -4.920346419227165, -2.4779264310825573, 2.923001315095547, -0.28494696524923036, 0.7432508387047566, 0.535732399700767], "ratio96": 1.5483427824386224, "ratio_final": 1.5768934243014407, "diag": {"ratio": 1.5768934243014407, "nonnormal": 0.9801960261134606, "eig_gaps": [0.5859229767047134, 0.6014950422817369, 0.7382617410942227], "discness": 0.11000412595186242, "peaks": 3, "deg": 3, "diamW": 2.7893237924236307}, "class": "stall", "probe_best": 1.5798006025023086} +{"family": "companion", "seed": 504002, "x": [0.13141908225047094, 0.5047987259507355, 0.29658858316490755, -0.41687424220017116, -0.5506997716056026, -1.9835134020835696, 1.5597868784088615, 0.16194955206448536, 5.5647024766702e-06, -1.273270165898631e-05, 1.6863752470423533e-05, 8.654171542564112e-06, -0.020253105774423685, 0.21654849374475144, -0.14667118719202327, 0.35968735268161456, 0.29554001836849797, -0.07799626145338417, 1.1031540098269728, -0.02311857945943209], "ratio96": 0.9999999999739742, "ratio_final": 0.9999999999848634, "diag": {"ratio": 0.9999999999848634, "nonnormal": 0.5601991160367072, "eig_gaps": [0.30693788359577445, 0.5838398438471112, 0.8457623830771873], "discness": 0.4109165248449733, "peaks": 1, "deg": 3, "diamW": 3.0506358603388426}, "class": "one-exact", "probe_best": 0.9999999999859668} +{"family": "peaks", "seed": 505002, "x": [-1.2473312389132412, 0.03483507952341347, 0.5960233822065992, -0.022482513288902994, 0.6411452621046123, -0.9466044008936096, 3.1029711060125162, -0.47929223035202656, -0.2865973169105285, 0.6008059211668431, 3.5236827012701433, 0.10375408036711858, -1.8921514392618834, -1.203957732127999, -1.1613934911893078, 1.7721158091576523, 2.438359456192721, 3.322899959800054, 0.0746165113870571, -0.24761317777520375], "ratio96": 1.9813651608710636, "ratio_final": 1.981347870661632, "diag": {"ratio": 1.981347870661632, "nonnormal": 0.7110121093359228, "eig_gaps": [0.17558135410788492, 0.3499861046472854, 0.40388749292477644], "discness": 0.03605227599254915, "peaks": 3, "deg": 3, "diamW": 5.269482139423678}, "class": "near2-other", "probe_best": 1.981347870661632} +{"family": "real", "seed": 506002, "x": [-1.9744829140393643, -0.8399882907828158, 2.259738070149643, 1.1104163517102132, 2.5011186393930487, -0.15750267038195365, 0.752056089505021, 1.9131115974699584, 1.0227611939658399, -1.3263516109788505, 0.2665817986910771, 0.07320527625896234, 0.26833747542995623, 1.1868505946426744, -13.705605780231839, -0.5019332022684981, 1.10471242329622, 0.2240006052298853, -1.9186357631211934, 0.7505899783125412], "ratio96": 1.2827768318740689, "ratio_final": 1.287282698565298, "diag": {"ratio": 1.287282698565298, "nonnormal": 0.8173469118523187, "eig_gaps": [0.2431218863724856, 0.8527952394211679, 0.8781297236251565], "discness": 0.22252116856972248, "peaks": 2, "deg": 3, "diamW": 5.308822693759193}, "class": "stall", "probe_best": 1.289471372725545} +{"family": "random", "seed": 500003, "x": [-0.715308304058238, -0.7231779561539313, -0.13033034729365361, 0.9093040880527814, -0.8861335077932107, -1.2060453053955138, 0.24806495176617993, 0.14425334429071718, 2.9697399364565578, 0.4663671992251304, 1.338263799926328, 2.9527004148812477, -3.030125335789352, -0.9106344249467557, -2.946471786311541, 0.8084999847297709, 0.03516201708643593, -0.15964510868482729, -0.029539834404597815, -0.00752653130309763], "ratio96": 1.9349121215729068, "ratio_final": 1.9359821267482429, "diag": {"ratio": 1.9359821267482429, "nonnormal": 1.2768319404358481, "eig_gaps": [0.10217253315737142, 0.3459243000770401, 0.448096186196038], "discness": 0.06393708645344323, "peaks": 4, "deg": 3, "diamW": 5.013024843525888}, "class": "stall", "probe_best": 1.9867437609330254} +{"family": "jordan3", "seed": 501003, "x": [0.5577604597064894, 0.26501883573393803, -0.08759005459389191, -0.282740039493834, -0.26140095148288356, -0.3952171977108754, -0.5456429414666236, 1.277780618277661, 0.02202211475810789, 0.04487673316807245, 1.518859221133376, -0.3977283305738564, 0.04906980580102918, -0.15827825384978667, -0.3839636005093281, -0.08795002044246919, 0.9740185719077649, 0.8416296161948017, 0.2438031105306604, -0.0743518230313806], "ratio96": 1.958941064105588, "ratio_final": 1.9594882050402616, "diag": {"ratio": 1.9594882050402616, "nonnormal": 0.7037578899594588, "eig_gaps": [0.08728840115055163, 0.3568921065045617, 0.44359345622539814], "discness": 0.05179686475545965, "peaks": 3, "deg": 3, "diamW": 2.371790518197552}, "class": "stall", "probe_best": 1.9669742586873553} +{"family": "jordan2", "seed": 502003, "x": [0.14662944646952766, -0.10673327840047382, -0.23098812019344794, -0.4187386068919553, -0.2691805248312558, -1.0577093654039518, 1.4398323586109718, 0.8273259996840779, -0.04598085778348075, 0.14244362953644502, -0.07644354790311306, 0.04139629027584234, -0.0874244929077668, 0.20153109259657187, 1.0736231761632362, 0.34167306849291046, 0.10962787486450135, -0.07202238719092877, -0.006625271898595066, 0.00619996143280397], "ratio96": 1.9716376557634991, "ratio_final": 1.972273657766895, "diag": {"ratio": 1.972273657766895, "nonnormal": 0.9669712478315908, "eig_gaps": [0.2816454424314092, 0.3680481741268153, 0.5967716990950583], "discness": 0.020767633998951637, "peaks": 3, "deg": 3, "diamW": 1.7392048090791261}, "class": "near2-other", "probe_best": 1.972273657766895} +{"family": "nearnormal", "seed": 503003, "x": [-0.5622071713189568, 1.0484692032467515, -0.1258844537520527, -0.787777972177629, -0.5303588755019489, 0.17556588943707369, 0.6171553996370316, 1.391718497682322, 0.9119404244102896, -0.9214545443532596, -2.199831896952622, 0.8786679420236467, 2.1638443276502493, 0.21786843988738036, 1.1440069021816004, -3.6348165084616513, 0.5384246951769596, -0.3947311591286482, 0.06062071609798446, -0.04663732821447272], "ratio96": 1.8524578282532453, "ratio_final": 1.8592570938725468, "diag": {"ratio": 1.8592570938725468, "nonnormal": 1.1519147239720757, "eig_gaps": [0.2585188090777314, 0.3092253239668432, 0.5585925460925428], "discness": 0.046518830997230164, "peaks": 3, "deg": 3, "diamW": 3.378802973247606}, "class": "stall", "probe_best": 1.8614515181593987} +{"family": "companion", "seed": 504003, "x": [-1.2644211293910106, -1.3016927261452884, 0.3709218147734667, -0.6162515363404695, -1.0786114985851927, -1.7153717828522685, -0.1573649704175326, -0.40518164901645926, -0.53642279689854, 0.5559109827149421, 0.21080502866483053, 0.18923319665354532, -1.2854640129119081, 0.4453704584437431, 0.6454823698499458, -0.02641882628190919, 0.2021726082130687, 0.6424159478123455, 0.5594934163598275, 0.28867723087423935], "ratio96": 1.5764737279162881, "ratio_final": 1.5720460083945795, "diag": {"ratio": 1.5720460083945795, "nonnormal": 0.18232005062994341, "eig_gaps": [0.20363738589590027, 0.7962339652765743, 0.8168636455704699], "discness": 223377024353.84473, "peaks": 1, "deg": 3, "diamW": 2.2269616405524753}, "class": "stall", "probe_best": 1.6180540869251172} +{"family": "peaks", "seed": 505003, "x": [-1.3173909570640232, -0.04806319439464175, -0.22161130503647225, -0.31929287991635824, 0.3429392048734153, -0.024468210005288475, 1.6052861631331583, 0.2980863017313476, -0.16542978022622318, -0.29254571460665246, 0.012214540847202094, 0.01161458023389096, -2.2306113217603913, -1.4209412021671444, -2.4100672453262035, -0.40289975456735183, 1.2381411773904203, 0.4266890635572962, 0.4564300709791528, 0.22879188838872633], "ratio96": 1.8534265506140137, "ratio_final": 1.8616595183047597, "diag": {"ratio": 1.8616595183047597, "nonnormal": 0.9984655437224198, "eig_gaps": [0.29639294655850634, 0.525331794413327, 0.7727453348111146], "discness": 0.3756073715048909, "peaks": 3, "deg": 3, "diamW": 2.1488292883344355}, "class": "stall", "probe_best": 1.8636279826281836} +{"family": "real", "seed": 506003, "x": [-1.2080906630373835, -0.7811207897958533, -1.1195841476372272, 1.0345494421196282, 2.6276581436218707, -0.850597606771537, -3.443115191884935, 0.7570862126481509, 0.959278671554517, 1.7839468803866219, -2.3346581999626213, -0.47946775551561716, -4.2014664811727105, 2.169204016777977, -1.0972488931203304, 1.0788164767425936, 1.8485943836818886, 0.5517116761443153, 0.07764830133492373, -0.07342991020672082], "ratio96": 1.8642393755887576, "ratio_final": 1.8660990242562745, "diag": {"ratio": 1.8660990242562745, "nonnormal": 0.7317393819973328, "eig_gaps": [0.2782673622561853, 0.5872612172160051, 0.6421138306027048], "discness": 0.11494911667945612, "peaks": 3, "deg": 3, "diamW": 6.532660183356133}, "class": "stall", "probe_best": 1.9492277504469258} +{"family": "random", "seed": 500004, "x": [0.09443646283812443, 1.765449821709887, 1.2320396193054757, -0.18546029232021768, 2.903672737728493, 2.1854671867621223, 0.11541977117164072, -0.09177120863394156, 2.3271303780865797, -0.2644608679375053, -2.9303502566095196, 1.6644332309452257, -1.4087538164228475, -4.519675698247865, 0.4777348511164487, 0.43174499570285146, 0.5894737798837215, -0.0267366511770797, -0.025787958336093042, 0.039865754808739426], "ratio96": 1.7181772332567675, "ratio_final": 1.7182643773409472, "diag": {"ratio": 1.7182643773409472, "nonnormal": 0.8294132665945488, "eig_gaps": [0.4552150302725093, 0.5725485396982, 0.5847460673880731], "discness": 0.3983510821991655, "peaks": 3, "deg": 3, "diamW": 4.961084637317503}, "class": "stall", "probe_best": 1.8918729946746151} +{"family": "jordan3", "seed": 501004, "x": [-0.1746341831548326, 0.3145458174334239, 0.10659994484091848, -0.1163687649503421, -0.19041755665805288, -0.28126637967842144, 0.9470418251354056, -0.3346179619680207, 0.08270956933747875, 0.037129517538420406, 0.8739519035789782, -0.5988611587734088, 0.058549100054516044, -0.0006001775963409082, 0.18056774791411653, 0.07373792650645855, 0.9388517570738802, 0.05126126457356961, -0.2093053104656878, -0.13385437254405727], "ratio96": 1.9690107251209321, "ratio_final": 1.971277939419618, "diag": {"ratio": 1.971277939419618, "nonnormal": 0.7547627528563305, "eig_gaps": [0.21472318236310392, 0.3252360274008694, 0.3767192107898742], "discness": 0.02466560465157238, "peaks": 3, "deg": 3, "diamW": 1.5821365044431808}, "class": "near2-other", "probe_best": 1.971277939419618} +{"family": "jordan2", "seed": 502004, "x": [0.029770918540695426, 0.013557312220732073, -0.15977660775375163, -0.33363837757640225, 0.1098118399489057, 0.0767037826395883, 0.644499200718647, 0.4853901970700765, 0.3331325909300783, 0.5431415590401549, -0.0665658024729473, -0.466338091303627, 0.041145155195050576, 0.12457357806795005, 1.0924514807698684, -0.0017601876873711367, 0.05324737309374375, -0.1621131565020537, 0.005834214919578669, 0.04039951879206363], "ratio96": 1.9718541478748715, "ratio_final": 1.972615867427874, "diag": {"ratio": 1.972615867427874, "nonnormal": 1.3189958465037148, "eig_gaps": [0.08587348071681365, 0.333186219219635, 0.4135504553806763], "discness": 0.024966320158555027, "peaks": 3, "deg": 3, "diamW": 1.1872244424756242}, "class": "near2-double", "probe_best": 1.972615867427874} +{"family": "nearnormal", "seed": 503004, "x": [0.06095555551291205, -0.083736792978448, 0.9111889820095083, 0.33476389729345113, 0.9712318531766985, 0.46378077163990006, -0.000899235563300465, 0.0007690682897037891, -0.00033244121294242254, 0.00039600134347156054, -8.525181562321008e-05, -0.3690307400006697, 0.13086975159764064, -0.23143688097129234, 0.4045386371093611, -0.049515347313990765, -1.0978060106223673, -0.2770076888831402, 0.9856723264127631, 0.2778082087419983], "ratio96": 1.764633951521981, "ratio_final": 1.7991307931216087, "diag": {"ratio": 1.7991307931216087, "nonnormal": 0.09184018604464274, "eig_gaps": [0.11884779552100579, 0.7914455149893928, 0.8871573168118372], "discness": 148900207637.48236, "peaks": 4, "deg": 3, "diamW": 1.1973655190791341}, "class": "stall", "probe_best": 1.8018478600177794} +{"family": "companion", "seed": 504004, "x": [-1.365896881162123, -0.9432807423943181, -0.8452534096982203, 0.2563704112078332, 1.2659592152279493, -0.8762703518481754, 1.0612093126535482, 1.6302626157109519, 0.574227648050369, 0.9456953263625429, 1.6996663293408303, -1.7246278804079278, -1.0871843632180007, 3.0063428186073526, 5.667182605568151, -0.19741049956292933, 2.5384482576257392, -4.357006186738648, -0.575322904734467, 0.19208072731123826], "ratio96": 1.848448954940936, "ratio_final": 1.8727094280973273, "diag": {"ratio": 1.8727094280973273, "nonnormal": 0.7109458838730016, "eig_gaps": [0.3069856859548061, 0.5624061261852886, 0.6180070245209218], "discness": 0.0954617661388771, "peaks": 3, "deg": 3, "diamW": 4.259998573538774}, "class": "stall", "probe_best": 1.8734048549653892} +{"family": "peaks", "seed": 505004, "x": [0.27815110116216446, 0.26424729329956953, 0.49918653906930366, 0.23745530448622304, 1.3092797513235106, 0.08769151601423145, 0.099587768566181, -0.12892805755074208, 0.368868495019229, 0.10484222845517964, 0.8475634895113329, 0.8218501699120739, 2.786182809227646, 1.1593813720719077, -2.5159470710738248, 0.17098400000631025, -1.3419344062036527, -1.2513353950627608, 0.5570338213136496, 0.3848045969635858], "ratio96": 1.940438693852559, "ratio_final": 1.9439309034314967, "diag": {"ratio": 1.9439309034314967, "nonnormal": 0.6934884741200089, "eig_gaps": [0.15023467118501335, 0.5558705913065035, 0.7058766224959847], "discness": 40720464314.08041, "peaks": 4, "deg": 3, "diamW": 1.482036477554378}, "class": "stall", "probe_best": 1.9723373760825906} +{"family": "real", "seed": 506004, "x": [-0.19296662767299722, -1.1466083489249699, 1.6684144427296443, 1.1515756335479441, 2.6298097261134785, -0.20413891702813436, 1.7375960387690688, 0.3338091567189336, 0.5028708787442547, -0.0779469676873566, -2.177656808763828, 0.22206767952174902, -0.4365914809707139, -2.772007601634985, -2.886306284173396, 2.734238081708968, 1.2854400520018454, -1.4466176540843456, 0.04311411472130805, 0.3002295317268304], "ratio96": 1.8437084745447607, "ratio_final": 1.8437241614229833, "diag": {"ratio": 1.8437241614229833, "nonnormal": 0.5225718939628375, "eig_gaps": [0.3820069926373305, 0.6797587115936086, 0.6840170258638105], "discness": 0.3537683706612981, "peaks": 4, "deg": 3, "diamW": 4.350704780585473}, "class": "stall", "probe_best": 1.8693458842105706} +{"family": "random", "seed": 500005, "x": [-2.069087991696604, 1.846251733041055, -0.6461220354212477, 1.3069586145382845, 1.8690967294694711, 0.061165639939162894, 0.2999667520400482, 0.30004074115845636, 2.7346458905501414e-07, -0.2300046831896692, -1.7280800613736385, 1.1311797804746404, -2.3090807425558726, -0.39123693455312414, 1.922363775231074, -0.2737550967542935, 0.32342386830418157, -1.3695032017575492, -0.7046835200210926, -1.6594576786847801], "ratio96": 0.9999999999695706, "ratio_final": 1.0229491721467197, "diag": {"ratio": 1.0229491721467197, "nonnormal": 0.5757398622173365, "eig_gaps": [0.32369047129240713, 0.5970471356753759, 0.9197378236866923], "discness": 0.5493964662242448, "peaks": 1, "deg": 3, "diamW": 4.701195836329327}, "class": "stall", "probe_best": 1.0363183488474945} +{"family": "jordan3", "seed": 501005, "x": [0.4570628799091081, -0.44529633184941947, -0.0608843119631914, -0.41388497600736174, -0.5981879725053629, -0.35143321185530885, 1.4724286133004298, 0.3460775045853062, 0.11899381183506011, 0.08711187384368109, 1.3086284792302691, 0.425579182398219, -0.27846828497663934, 0.11378403269398368, 0.24281562826250874, 0.7973600507965855, 0.9425746239490878, -0.13947498728706625, -0.11503539414762576, 0.018230992363778453], "ratio96": 1.9910387394827864, "ratio_final": 1.991434563410379, "diag": {"ratio": 1.991434563410379, "nonnormal": 0.6510105005408897, "eig_gaps": [0.22547219616273434, 0.23504126099530165, 0.4603385189360222], "discness": 0.06555872874232109, "peaks": 4, "deg": 3, "diamW": 2.3013870931919884}, "class": "near2-other", "probe_best": 1.991434563410379} +{"family": "jordan2", "seed": 502005, "x": [0.15919421226264957, 0.04347563590027815, -0.1971257888099483, 0.0740641677200744, 0.19888345549799358, 0.09251555704917194, 0.6445481008248621, 0.4631113163716093, 0.2022965820124849, -0.9049485275262567, 0.09983030240558963, 0.5029770490281956, -0.06050894556624314, -0.1800735256268043, 2.5620678460187447, -1.0683125867008954, 0.045336270215217876, -0.16527779098743664, -0.11347188151143842, -0.015109274823685822], "ratio96": 1.9900613248531611, "ratio_final": 1.9900582366896558, "diag": {"ratio": 1.9900582366896558, "nonnormal": 1.3625083941209595, "eig_gaps": [0.0460751398593191, 0.2611869919164508, 0.2895297360875586], "discness": 0.014011156991762483, "peaks": 3, "deg": 3, "diamW": 1.369250953130337}, "class": "near2-double", "probe_best": 1.9900582366896558} +{"family": "nearnormal", "seed": 503005, "x": [-0.7592780891492801, 0.961520811570451, 1.0578117427317606, 0.9836206167029982, -1.991411636052115, -1.254921000573472, 0.9164382545946852, 0.04378724456273925, -0.7945787582888213, 1.4003280646601892, 0.3205448694452452, -0.8275076854905372, -0.18077706391441456, -1.1202837296346055, 0.5015972216151467, -0.6362871644273103, 1.938377974268537, -0.9515683811755578, 0.27345799424469786, -0.26604744593393936], "ratio96": 1.4123978846974081, "ratio_final": 1.5129410425167429, "diag": {"ratio": 1.5129410425167429, "nonnormal": 0.6216950295816026, "eig_gaps": [0.426300455559366, 0.5948927565566945, 0.8873794556545801], "discness": 0.38185934327894167, "peaks": 3, "deg": 3, "diamW": 4.262778034386288}, "class": "stall", "probe_best": 1.519486363591183} +{"family": "companion", "seed": 504005, "x": [1.5814236674850601, -0.8721879896223544, -1.0972020823466826, -0.46230097708718043, -0.11403948352747335, -0.9216309586478708, -0.44298277326437796, -2.0076884067079925, -0.24945458909342266, -0.7165680683500626, 1.9195274011876828, 0.8065171374852447, 3.049156693192973, 1.5951112509697427, 1.0010514159772717, -1.8422829357060242, -0.9365331942570867, -0.03465594031179179, 0.6223160897966367, -0.2751499163917813], "ratio96": 1.6185508319284885, "ratio_final": 1.6276224592666562, "diag": {"ratio": 1.6276224592666562, "nonnormal": 0.6703509043694417, "eig_gaps": [0.2662599682714382, 0.4161800529859599, 0.6648847349590461], "discness": 0.16449730865120551, "peaks": 3, "deg": 3, "diamW": 4.075601201234808}, "class": "stall", "probe_best": 1.6422561910157558} +{"family": "peaks", "seed": 505005, "x": [-0.7612558511373327, 0.9485626067958162, -0.2428870975528662, -0.6495226125549602, 0.5943327224549597, 0.8616030889176765, 1.8337344297505629, 0.6481842998416011, 0.1485285994921452, -0.2137414249253975, 1.7682802179326647, -0.321043210951741, -1.6746131966458149, 1.550421096824149, -3.3759773378753573, -3.0401967764682727, 1.191152161069493, -3.2718891288469067, -1.0170901910598675, -0.6620272776109222], "ratio96": 1.9117985491108143, "ratio_final": 1.9123656411555223, "diag": {"ratio": 1.9123656411555223, "nonnormal": 0.7457271605630227, "eig_gaps": [0.4319571663780295, 0.5342497945927659, 0.5493540280096324], "discness": 0.03086325421633432, "peaks": 4, "deg": 3, "diamW": 3.1446981195378285}, "class": "stall", "probe_best": 1.9130435328400517} +{"family": "real", "seed": 506005, "x": [-0.5340120983283243, 0.4255701872537909, -1.3075161397115775, 0.2103137030505705, -0.13958026211815006, 0.133790541829984, 1.1988638191140861, -0.03434047700854527, -0.5240926456713064, 0.4274002669151292, 0.754718722250387, -0.1511003311793701, -1.3047854871209248, 0.4531477395117032, -0.9769323102494223, 0.08457192779511816, 0.5605575174556912, 0.07816684378213559, 0.13138651462636897, 0.038345216910724775], "ratio96": 1.9133284137402589, "ratio_final": 1.913455138646116, "diag": {"ratio": 1.913455138646116, "nonnormal": 0.8287663979654507, "eig_gaps": [0.2617298706796341, 0.42831619207971355, 0.6243868649767325], "discness": 50903268374.86762, "peaks": 4, "deg": 3, "diamW": 1.874543103645356}, "class": "stall", "probe_best": 1.9242903787382113} +{"family": "random", "seed": 500006, "x": [-0.41204351132940115, -0.23741187869780533, 0.592671956696535, 0.5705249481514083, 0.3769118752944296, -0.04123990295187596, -0.6385099760625764, -0.0062821526399785286, 0.7228527309761854, -0.04738617696154368, 0.0836510598893102, 0.861971928623388, -0.6933115250375579, 0.3837853710985154, -0.43070744085808893, 1.0851426872305678, 0.32152530579968874, -3.2260321702965324, -1.1647415666738032, 1.0639066177465153], "ratio96": 1.7705055706620754, "ratio_final": 1.8827341824610222, "diag": {"ratio": 1.8827341824610222, "nonnormal": 0.7501048303278071, "eig_gaps": [0.35460385985594883, 0.4444062862065038, 0.7047656710825362], "discness": 0.16244895109505675, "peaks": 4, "deg": 3, "diamW": 1.8293586929726868}, "class": "stall", "probe_best": 1.9155427833519216} +{"family": "jordan3", "seed": 501006, "x": [-0.04099317945531747, -0.1079389890056528, 0.20226388418521069, -0.6741873204955526, -0.01435239544943711, 0.04609913872744986, 0.834863405984223, -0.07857395257133193, 0.5983213951230593, 0.18964257571410637, 0.5194033912620527, -0.28253441969230153, 0.26747345793938815, 1.0946541047294271, 2.113760511701488, -1.5044692306064833, 0.5337996308482178, 0.4312937095305877, 0.2530754550923772, -0.39923175784314635], "ratio96": 1.9545744982898716, "ratio_final": 1.9559317390935447, "diag": {"ratio": 1.9559317390935447, "nonnormal": 1.1725234387015397, "eig_gaps": [0.11331276250277836, 0.4467190977877997, 0.5452017624020895], "discness": 0.12088611434573038, "peaks": 3, "deg": 3, "diamW": 1.3795878092625207}, "class": "stall", "probe_best": 1.958474485919343} +{"family": "jordan2", "seed": 502006, "x": [-0.16770548867426338, -0.09385848358388077, 0.1433584941105634, 0.17909134342675526, -0.07200854615851088, 0.2557086653098921, 1.5420857331971387, 0.18378922458726507, -0.5756652218319607, -0.7021691789459082, -0.0034724147345558867, -0.15914449691567167, -0.1351298912452304, -0.24858173300943542, 2.605135825429908, -0.37075718537111907, 0.16033447389032324, -0.09334961144406909, 0.01671396835935704, 0.02277950442062541], "ratio96": 1.9876389785513804, "ratio_final": 1.9878657349927435, "diag": {"ratio": 1.9878657349927435, "nonnormal": 1.3836096000718354, "eig_gaps": [0.12313718823144582, 0.1952344034733779, 0.22292763615400507], "discness": 0.0087149157093932, "peaks": 3, "deg": 3, "diamW": 1.8563809963014617}, "class": "near2-double", "probe_best": 1.9878657349927435} +{"family": "nearnormal", "seed": 503006, "x": [-1.015929884400523, -0.4058456269200636, -1.1616772546706022, -0.3693554686800117, 0.15633061522902567, 0.19510581243919597, 0.3740584594215309, -0.48907632687741853, -0.012183374456353743, -0.0019162642343549473, -0.002967672357479615, -0.01239131186041461, -0.5835036135418304, 0.7799393213503123, 1.8937886546641065, -0.38636111535848155, 2.2997654186873944, -0.33052735239152076, 1.0956249129654576, 0.5964199725645171], "ratio96": 1.7655613360193627, "ratio_final": 1.603390187982936, "diag": {"ratio": 1.603390187982936, "nonnormal": 0.17682404839141916, "eig_gaps": [0.08899935189541358, 0.7803262100991092, 0.8493184988563877], "discness": 199309546080.66168, "peaks": 1, "deg": 3, "diamW": 1.6881683655365383}, "class": "stall", "probe_best": 1.7756817439180557} +{"family": "companion", "seed": 504006, "x": [-0.40707152080667064, -0.3323352698645867, 0.061761814900964135, -0.2859200946784225, -0.6989047737230032, -0.15138172324661517, 0.7115108309266689, 0.2691714017656131, 0.1178459243263385, -0.19627602161907687, 0.7409243915430457, 0.09316032883771098, 0.23852516001208546, -0.2395022537292576, -0.18732685495204637, -0.8501833797750606, -0.1517098589834508, -0.07159919596960171, 1.2676792022126369, 0.3638652184281414], "ratio96": 1.8188771907929755, "ratio_final": 1.8219698678474072, "diag": {"ratio": 1.8219698678474072, "nonnormal": 0.6531346143408411, "eig_gaps": [0.26012287991754646, 0.3568930923932111, 0.5851738321588176], "discness": 0.16717708696567973, "peaks": 4, "deg": 3, "diamW": 1.32007407241262}, "class": "stall", "probe_best": 1.8496531034016321} +{"family": "peaks", "seed": 505006, "x": [-0.6256433503264893, 0.7712813364613571, -0.38785147574244483, 0.4213983942233402, 0.15874856559299771, 0.2270637541056873, -0.017552209990751758, 0.8595389189655733, 0.5312314995011365, -1.6787254496472575, -0.3450772730947873, 0.5400167769849096, 1.578791288436448, 1.298092735708945, -0.024056010274912547, 4.3493681641162745, -0.18975044869812174, -0.5214739881650416, 0.18959880843139426, -0.07810949504799372], "ratio96": 1.9405249948072203, "ratio_final": 1.9423007722174221, "diag": {"ratio": 1.9423007722174221, "nonnormal": 1.1315153197544892, "eig_gaps": [0.18730863508019316, 0.25685787388847303, 0.4227084985817874], "discness": 0.05626686158109703, "peaks": 3, "deg": 3, "diamW": 2.2585197143533873}, "class": "stall", "probe_best": 1.946318060055732} +{"family": "real", "seed": 506006, "x": [-1.2918488970062563, -0.10438299700344839, -0.47265524471854303, 0.015220156592217586, 1.5700104483404893, 0.09490618797059618, -0.9908854020488214, 0.16416240919722266, 0.2900271651558012, 0.29000000279660254, 0.24999895450772702, 0.26000191663941574, -0.39308795296612287, 0.10992210825524883, -1.148670624486136, -0.246928489559761, 0.19784913535130422, 0.13411384787682165, -1.0549001590614566, -0.10930324545228666], "ratio96": 0.9999999999680416, "ratio_final": 1.107756656457773, "diag": {"ratio": 1.107756656457773, "nonnormal": 0.5156420173666064, "eig_gaps": [0.26255102564919564, 0.6482977493695948, 0.9097995945232918], "discness": 0.3248587741345354, "peaks": 2, "deg": 3, "diamW": 3.153210715642043}, "class": "stall", "probe_best": 1.1191449637896473} +{"family": "random", "seed": 500007, "x": [0.5982539367085842, 0.28087590533948237, -0.2906140044868054, 1.5287685463985716, 0.35830386617967536, 1.3045691643083366, -1.7640176945609476, -0.5295464883022528, 0.9865775741125192, 1.197752756906958, 0.3362129228298452, -2.795257184889161, -0.1467066315824433, -4.601160240095304, 1.2729368957596483, 0.5106827039219279, 0.514489719657443, -0.7490674780614797, 0.14212302564008333, 0.004892357881707492], "ratio96": 1.677985221813004, "ratio_final": 1.6783193414620063, "diag": {"ratio": 1.6783193414620063, "nonnormal": 0.8823679468612136, "eig_gaps": [0.17438630343903028, 0.26706697661712436, 0.3891548754457549], "discness": 0.25692324662435945, "peaks": 3, "deg": 3, "diamW": 3.936986112935207}, "class": "stall", "probe_best": 1.877236003791767} +{"family": "jordan3", "seed": 501007, "x": [0.7182254905448828, 0.16030134838626608, 0.9122742906885475, 0.2823604819306559, 1.0624700291923084, 0.37259406962090824, 0.16406860333724005, 0.3820887198250327, -0.48569723101919426, 0.5877173335259569, 0.22899519983624353, -0.19258469839446749, -1.3989498339000828, -0.32940194759614383, 0.8227203132910446, 0.06740023071648579, 1.185550498161223, -0.4264692450052471, -0.3448310032757447, 0.30944420960369423], "ratio96": 1.9607612844343567, "ratio_final": 1.9609974357998128, "diag": {"ratio": 1.9609974357998128, "nonnormal": 0.34689406879099965, "eig_gaps": [0.1767359254835867, 0.23123301778026764, 0.40794791189414875], "discness": 15618905373.987396, "peaks": 4, "deg": 3, "diamW": 0.9914034677881135}, "class": "stall", "probe_best": 1.962838444156481} +{"family": "jordan2", "seed": 502007, "x": [0.26080568341964794, -0.2491262742854301, 0.30008698970607117, -0.4355062912243324, 0.24838728265124332, -0.4496276832745092, 0.26041543429098024, 0.06565815412669812, -0.1675685667862731, 0.5650797977444564, 0.044818042902740274, 0.07930371125220992, -0.668099895448035, 0.3882987212505953, 1.3981420171935612, 0.4203852264196202, -0.3321939690202995, 0.5506009384546662, 0.34489907491191385, 0.14488969860468676], "ratio96": 1.986723249553613, "ratio_final": 1.98673415332121, "diag": {"ratio": 1.98673415332121, "nonnormal": 0.5781103498401133, "eig_gaps": [0.07788017843962519, 0.27679030219565603, 0.29191935740306485], "discness": 5698634131.2727165, "peaks": 3, "deg": 3, "diamW": 0.6881544964944972}, "class": "near2-double", "probe_best": 1.98673415332121} +{"family": "nearnormal", "seed": 503007, "x": [-0.31799881471235397, 1.2654732962048028, -1.4693660359741814, 0.4178021774339271, -0.4063411776823115, 0.5368709136224183, 0.37998015139571395, 0.22001915593250287, 0.8379347427050269, -0.42587196533156857, 0.25999597367673916, 0.14998582107063693, -0.2443141686586502, -0.26419516019293754, 0.0772714960614027, -2.730030057576001, 1.6340257235466662, -2.564719508731673, 0.8387196304338773, 0.18069352132027516], "ratio96": 0.999999999971606, "ratio_final": 1.311502965762841, "diag": {"ratio": 1.311502965762841, "nonnormal": 0.35480550185754545, "eig_gaps": [0.40698431166340937, 0.5931558081843561, 0.7928278410905608], "discness": 116307873369.65643, "peaks": 2, "deg": 3, "diamW": 1.8033583498409518}, "class": "stall", "probe_best": 1.3159003467212247} +{"family": "companion", "seed": 504007, "x": [1.2529431646827507, -1.6422366089660185, 1.3171001441043773, -0.7645633521037738, -1.1030415833394314, 0.43659455750908294, 1.532872301315984, -1.0565037872392107, -1.250121774494891, -0.021375109633892803, 1.8992767229540413, 1.679916347704791, -0.18219845606579294, -0.08278840727167826, -0.32964545327125966, 2.1821154869598542, 0.8146976460219377, -0.20727144490442367, 0.2375512684639946, -0.05634312301504897], "ratio96": 1.8547129945369247, "ratio_final": 1.8730875103984932, "diag": {"ratio": 1.8730875103984932, "nonnormal": 0.7296587493399895, "eig_gaps": [0.18889282047723752, 0.5799396629357269, 0.6744230432901922], "discness": 0.14664118205353546, "peaks": 4, "deg": 3, "diamW": 4.658806162719001}, "class": "stall", "probe_best": 1.9397555784479596} +{"family": "peaks", "seed": 505007, "x": [-1.2706054051772044, -0.7507613150189849, 0.7803735041622817, 0.1664112393451057, 0.8932209831666835, 0.7370352125749706, 0.938765421948411, -0.039216441478698315, -0.39941338276714944, -0.10461853194911486, 1.68043286769929, -0.001469699185500082, -2.1603587608595274, -2.003173835141376, 0.699606590327484, 0.20822967266532855, 2.9062428915249185, -0.8878581035449749, -0.19968380806051345, 0.4448105416265147], "ratio96": 1.8582262016006263, "ratio_final": 1.8600022210979095, "diag": {"ratio": 1.8600022210979095, "nonnormal": 0.6062565026936417, "eig_gaps": [0.16890257717908858, 0.6523838562616526, 0.7625076883412438], "discness": 0.21605775369531335, "peaks": 5, "deg": 3, "diamW": 3.443851543881442}, "class": "stall", "probe_best": 1.8766052435850527} +{"family": "real", "seed": 506007, "x": [-0.7372805674436755, -0.1848465849047567, 0.7924225063417616, -0.15553906753021496, -1.4563507622507017, -0.048337109019312416, -4.009466721859688, -0.6036314111396238, -1.861748219024578, -2.1004308114021457, 1.4157142137767416, 0.7291437045249425, -0.3835899512498022, 0.29435297383769843, 1.9518797489724768, 0.9262601437816407, 0.03646633522872182, 0.002670816322623216, -0.007439743793897023, -0.0017536729146357278], "ratio96": 1.9894433643156342, "ratio_final": 1.9896822883667182, "diag": {"ratio": 1.9896822883667182, "nonnormal": 1.3124480003617534, "eig_gaps": [0.1342476160960317, 0.280629870399024, 0.4129387631330127], "discness": 0.020500337106828564, "peaks": 3, "deg": 3, "diamW": 5.451963458496321}, "class": "near2-double", "probe_best": 1.9896822883667182} +{"family": "random", "seed": 500008, "x": [-0.03457113233652699, 0.16945064901653617, -0.47024851474979135, 0.38306110816996564, -0.7223470818466753, 0.5933421152358827, 0.24831916881218624, 0.7631290120039853, -0.11653556870476323, 0.10915105826545674, 0.46231100818104187, 0.4681024148912718, 0.049292866210784665, -0.648468430538026, 1.6556312856198325, -1.2072988757047924, 0.37537438961766123, 0.5278892364870751, -1.2759196990172863, -0.6157660690768139], "ratio96": 1.9680422644981976, "ratio_final": 1.9684583081604405, "diag": {"ratio": 1.9684583081604405, "nonnormal": 0.4475722833311422, "eig_gaps": [0.24941425495278938, 0.3686487211061803, 0.6138073650890158], "discness": 0.5486714087413636, "peaks": 5, "deg": 3, "diamW": 1.316228455665601}, "class": "stall", "probe_best": 1.9748685244028554} +{"family": "jordan3", "seed": 501008, "x": [-0.22967552929408996, -0.6553382113902071, 0.09591273175143344, -0.21016545132886433, -0.447378708905801, 0.23592161109636045, 1.4754058355895234, -0.1659840238350465, -0.12602318140767338, -0.13202976062411786, 1.4817472393081825, -0.04360133906567049, 0.15630587191182743, 0.16240293612136097, 0.6677018847472302, 0.32273319396713585, 1.08417700664884, -0.3486996735121546, -0.13055511072750237, 0.11115942711080645], "ratio96": 1.9778554304637277, "ratio_final": 1.9779558369327876, "diag": {"ratio": 1.9779558369327876, "nonnormal": 0.6907212665393927, "eig_gaps": [0.23980163765553178, 0.3056438368148927, 0.39890624417229414], "discness": 0.03030247786563593, "peaks": 3, "deg": 3, "diamW": 2.299946981333288}, "class": "near2-other", "probe_best": 1.9779558369327876} +{"family": "jordan2", "seed": 502008, "x": [0.41367615487311693, -0.7052270593640513, 0.08829083198288962, -0.10135507223690005, 0.34829972436371504, -0.37789811148204056, 0.9632522689023822, 0.2764422942706417, -0.3117553644247148, 0.2191684404669858, -0.002824586931575168, -0.12219485459342572, -0.11564460031266272, 0.5066266995514819, 0.9340751361273734, -0.3449877309284852, 0.27769916620764024, 0.24304319334424207, 0.06224397839123205, -0.2736128288442259], "ratio96": 1.9530161222853863, "ratio_final": 1.9540779197750053, "diag": {"ratio": 1.9540779197750053, "nonnormal": 0.9208109449412336, "eig_gaps": [0.26102842657445924, 0.29683313351057794, 0.536421763741768], "discness": 0.3103013675079361, "peaks": 4, "deg": 3, "diamW": 1.2787642696599515}, "class": "stall", "probe_best": 1.9666374147244774} +{"family": "nearnormal", "seed": 503008, "x": [-0.03363069120592657, 0.8636346513180722, -0.0938011905603858, 1.1289404598312123, 0.07073659051785171, 0.6741977660886719, 0.4453939021242344, 0.5412399143123099, 0.09157780176287447, -0.09929145938264042, 0.10527323115379554, 0.048135467754098854, 0.3140269723016823, -0.854409041922854, -0.4943593871395099, 0.49222076268449766, -0.28276890888993655, -1.6652245354557493, 0.40719749894289303, 0.06884434114936022], "ratio96": 1.9353014219635578, "ratio_final": 1.9353272570811622, "diag": {"ratio": 1.9353272570811622, "nonnormal": 0.2613718356286785, "eig_gaps": [0.2773329546176574, 0.3488309310596085, 0.6200946181990649], "discness": 10073639129.064056, "peaks": 3, "deg": 3, "diamW": 0.779872043528181}, "class": "stall", "probe_best": 1.9493635439006967} +{"family": "companion", "seed": 504008, "x": [1.387211197120965, 0.806934086592578, 1.5698213100156604, 1.0240601348635172, 1.419219489927798, 0.8523088880075473, 0.04814582394035173, 0.2513965692415583, 0.2707237505686544, 0.13060816550858242, 0.5374980755144669, 0.12635263712353365, -0.4531184083505695, -7.370400403896171, 3.180732231244413, 0.21492749369755793, 0.15289407294122048, 0.29592472855422336, 0.5538826064788522, 0.5287440988515064], "ratio96": 1.9125944932253118, "ratio_final": 1.9146148700527859, "diag": {"ratio": 1.9146148700527859, "nonnormal": 0.07064830910752298, "eig_gaps": [0.07887069942619285, 0.3244517001282228, 0.4029692912045008], "discness": 5489290122.492132, "peaks": 3, "deg": 3, "diamW": 0.7040434819590969}, "class": "stall", "probe_best": 1.9173084861289078} +{"family": "peaks", "seed": 505008, "x": [0.5596459828032446, 0.44411317547688045, 0.5533300193677884, 0.0169462314332669, 1.2114364357569096, 0.019783570030853626, 0.010562561074090112, 0.08394017046089772, -0.03613876029336126, 0.2015246018398989, 1.0809734299510376, -0.7137323306054711, 3.0389462166199923, -0.4117799292534132, -2.798685527084828, 0.41308682599982816, -1.0742999886223332, 0.25874105128285885, 0.4152775474949746, -0.09139246136532], "ratio96": 1.9800725476913412, "ratio_final": 1.9802751031818326, "diag": {"ratio": 1.9802751031818326, "nonnormal": 0.6776918652028181, "eig_gaps": [0.2911664210120919, 0.4485350076101817, 0.5300698427835712], "discness": 26694510731.97992, "peaks": 4, "deg": 3, "diamW": 1.4672489807915494}, "class": "near2-other", "probe_best": 1.9802751031818326} +{"family": "real", "seed": 506008, "x": [-0.28458217987588463, 0.09780222291061479, -0.7126645796094393, 0.2413941537692197, 0.42375745146955834, -0.04810111770192084, -1.1838661829171668, 0.07822944898160443, 0.6755570565222454, 0.1543722884672247, -1.148010560031732, 0.051340400228365925, -0.10880549570683819, 0.01968204432621616, 0.3304377392629603, -0.1416025141469725, 0.8057197129020177, 0.07980509790805379, -0.06616076915753935, -0.07685133089051241], "ratio96": 1.9712091093854216, "ratio_final": 1.973556833656473, "diag": {"ratio": 1.973556833656473, "nonnormal": 0.6998178361065162, "eig_gaps": [0.21136005402167093, 0.3385379257237883, 0.5489536367057761], "discness": 0.07329277387037898, "peaks": 4, "deg": 3, "diamW": 2.13627491600623}, "class": "near2-other", "probe_best": 1.973556833656473} +{"family": "random", "seed": 500009, "x": [0.17283796570937904, 0.000932560475767798, 4.094841522668904, 2.0984402361351027, -0.9334244965510919, -0.28880150583732256, 2.531047066902074e-07, 9.033371110550715e-07, 0.7805629224037461, -1.4103742411905227, -3.006570006865509e-07, 2.6345838054169797e-06, 0.759265398313272, 0.9139806879430488, -0.6246164842289371, 0.33649451374968065, 0.46010058333015974, -0.052576551724925105, 0.7002803242832355, 1.0455120650446306], "ratio96": 0.9999999999724575, "ratio_final": 0.9999999999999187, "diag": {"ratio": 0.9999999999999187, "nonnormal": 0.18201175734948313, "eig_gaps": [0.19100856090928922, 0.7428816982398109, 0.9297063617270538], "discness": 1.3998786183233334, "peaks": 10, "deg": 3, "diamW": 5.987032304476571}, "class": "one-exact", "probe_best": 0.9999999999999215} +{"family": "jordan3", "seed": 501009, "x": [0.2821294326834504, 0.36895800863397477, -0.5650967479501139, -0.4809262717654036, -0.23654366878070246, -0.3016783921359869, 1.5234350749463461, -0.09508698764993168, -0.01850288102992754, -0.2355500170340836, 1.679901544510132, -0.5533669827441372, 0.11881582558146447, -0.1247943763454511, 0.20713268358299985, 0.36770117337767627, 1.19178189612852, 0.42610651674449096, 0.18633425420667798, 0.010331430572217908], "ratio96": 1.9853044653358427, "ratio_final": 1.9852625351617286, "diag": {"ratio": 1.9852625351617286, "nonnormal": 0.7188103264564445, "eig_gaps": [0.14166066587090936, 0.320894627625591, 0.4542151540130005], "discness": 0.05541818691616323, "peaks": 4, "deg": 3, "diamW": 2.6420073763429883}, "class": "near2-double", "probe_best": 1.9852625351617286} +{"family": "jordan2", "seed": 502009, "x": [0.026712999691534373, -0.017608943440771047, 0.04533415958651206, 0.39481666014206296, 0.021963244789375502, -0.09676162876056095, 0.8263693109113025, 0.2581495327098104, 0.2258348478589608, -0.22274932996377733, 0.23979487022154578, 0.13891871288793012, -0.02316074091573289, -0.21040667267026086, 1.1209809286783665, 0.19513514847400099, 0.09865750619618785, 0.04023938231473487, 0.12895462542856467, 0.06926039638386904], "ratio96": 1.9724903402031084, "ratio_final": 1.9773480551859004, "diag": {"ratio": 1.9773480551859004, "nonnormal": 1.271655869474597, "eig_gaps": [0.07693685862047848, 0.4005678689049317, 0.4774976438989443], "discness": 0.041011149981340894, "peaks": 4, "deg": 3, "diamW": 1.0306512267767316}, "class": "near2-double", "probe_best": 1.9773480551859004} +{"family": "nearnormal", "seed": 503009, "x": [-1.240817047639691, 1.7307599538166303, -1.991417356877642, 0.30037302785604053, 1.0987906982001805, 0.789919369628617, 0.2799974671855091, 0.2800100456146745, -0.38649321769613315, -0.4065612671325171, -0.02000845039108219, -0.17999054185869096, -2.1433252735345816, 0.4327651996219883, -1.9427293495394085, -4.100122259006257, 1.7107713304483945, -3.3911718221899374, 0.5321150702561429, 0.6156685682881737], "ratio96": 0.9999999999773331, "ratio_final": 1.176328823444026, "diag": {"ratio": 1.176328823444026, "nonnormal": 0.21560830750260926, "eig_gaps": [0.5033110836659874, 0.7857024052558023, 0.9748451109029588], "discness": 1.7041508459188444, "peaks": 2, "deg": 3, "diamW": 3.2094784626218775}, "class": "stall", "probe_best": 1.1808343442158362} +{"family": "companion", "seed": 504009, "x": [0.5713728664479059, -0.4847888733538984, 0.10566944107922101, -0.6964589370727375, 0.3053996760380849, -0.6218330471296867, 0.6880471570791842, -0.3319955478023392, -0.23386530412370654, -0.15153621120772764, 0.7892353558418967, -0.34227455440219134, -0.8649491859269002, 0.1170281344246965, 0.8569729830503166, 1.5056154621269964, 0.6344866166543877, 0.05380878648622285, 0.9928131466396728, -0.4775055028321879], "ratio96": 1.934878344455013, "ratio_final": 1.9352691803344193, "diag": {"ratio": 1.9352691803344193, "nonnormal": 0.4278927248596357, "eig_gaps": [0.1657354153421815, 0.23257437066057263, 0.39763391476203364], "discness": 22675913627.53026, "peaks": 3, "deg": 3, "diamW": 1.286486045668693}, "class": "stall", "probe_best": 1.9408341519686292} +{"family": "peaks", "seed": 505009, "x": [0.5614319949003304, -0.2948969282292919, 0.6998432198189439, -0.0984344722748367, 1.2142675173130146, -0.24424003445680253, -0.07937631299600059, 0.10446992353135731, 0.47145925098837377, 0.6092599200426175, 0.5658669475234552, -0.6160005820088137, 1.6944775255460751, -2.524344360043127, -2.0408678839332053, 1.425438903285118, -0.6424777783561826, 1.1432053664846555, 0.343505227658069, -0.3047366146381639], "ratio96": 1.9696616946033516, "ratio_final": 1.9697441665215818, "diag": {"ratio": 1.9697441665215818, "nonnormal": 0.547950833538861, "eig_gaps": [0.18979424057069688, 0.42226815636816234, 0.517124272443979], "discness": 18751508051.343136, "peaks": 4, "deg": 3, "diamW": 1.266229363716848}, "class": "stall", "probe_best": 1.976709313592914} +{"family": "real", "seed": 506009, "x": [-1.5032556651800535, 0.11389351538625425, 0.20704136140636978, 0.32239969964942733, -1.7506050192922924, 0.04623322560446694, 0.0007862150973544763, -0.00030156154302410316, 0.9481511715700149, -0.15296831108148917, 3.6438566880762207e-06, -0.0008522075254251533, -1.5041046737727004, 0.1267412717502434, -0.09075669229494204, -0.1425856571697962, -0.3183260692135189, -0.04646083745178978, -0.4352097620856077, -0.000312922089239914], "ratio96": 1.7512202359166855, "ratio_final": 1.7520955385073727, "diag": {"ratio": 1.7520955385073727, "nonnormal": 0.21067588593856093, "eig_gaps": [0.10927789569813624, 0.7342227686392612, 0.8424922672857671], "discness": 265281436732.8345, "peaks": 3, "deg": 3, "diamW": 2.346644579239637}, "class": "stall", "probe_best": 1.7709762437582186} +{"family": "random", "seed": 500010, "x": [-1.5095708789915676, 0.7394727232674816, 1.3099777971542839, -1.7386103798494057, -1.450378615302026, -1.7322112575725688, -2.873906860112511, -0.6776298407961584, -1.0680489412469576, 0.778068643343907, 1.392028999591648, 1.7234568350734734, 0.43788511827129317, -1.0309509104694448, -1.2282113646994133, 1.0333978880829524, 0.8864883518651041, 0.8895676117093843, -1.0232162024983886, -0.6042582550113238], "ratio96": 1.4465717235015545, "ratio_final": 1.45736747551724, "diag": {"ratio": 1.45736747551724, "nonnormal": 0.746666731857822, "eig_gaps": [0.4550946489282947, 0.5081016589498918, 0.6909575218661724], "discness": 0.11690930315995035, "peaks": 3, "deg": 3, "diamW": 5.432699896017127}, "class": "stall", "probe_best": 1.5734798390495512} +{"family": "jordan3", "seed": 501010, "x": [-0.1144253048693335, -0.18924399227146765, 0.3437064990649983, 0.5279498071967754, -0.5986328974211959, -0.6152520348467372, 2.069236341559678, -0.9737303289213511, -0.10551299240256229, 0.2872178382139535, 2.040411182727972, -0.23444894698092525, -0.08677404325123467, -0.4081438909213334, 0.7872413684801476, 0.377018112658817, 2.2217083587688857, -0.5781350325434547, -0.04121406317150289, 0.248897187737942], "ratio96": 1.9913879621825537, "ratio_final": 1.9915603286186754, "diag": {"ratio": 1.9915603286186754, "nonnormal": 0.7398884071167791, "eig_gaps": [0.18942661696375726, 0.24995999116958895, 0.43514549689129317], "discness": 0.0357252621042379, "peaks": 4, "deg": 3, "diamW": 3.404664008414236}, "class": "near2-other", "probe_best": 1.9915603286186754} +{"family": "jordan2", "seed": 502010, "x": [-0.30389812614964806, -0.23728416263662755, -0.008809228705477512, -0.1664858529644982, -0.12716971213834521, 0.006296288836362605, 1.2184949328315169, -0.7565783931654816, -0.8595323707527062, 0.24925744286410662, -0.23902891414240857, -0.05244624487938482, 0.2766614715127791, 0.1615303548990042, 2.0985367212757944, -1.7043719050719144, 0.056236528964586155, -0.35477077672605817, -0.016757479097702158, -0.012395512122671856], "ratio96": 1.9725848769471406, "ratio_final": 1.9726171510423114, "diag": {"ratio": 1.9726171510423114, "nonnormal": 1.3388737146635046, "eig_gaps": [0.12041467179846696, 0.17302527701617804, 0.17447643263561038], "discness": 0.007913642179095997, "peaks": 3, "deg": 3, "diamW": 1.7392785010015854}, "class": "near2-double", "probe_best": 1.9726171510423114} +{"family": "nearnormal", "seed": 503010, "x": [0.22702796052230567, -1.8698423864234233, -0.29818029006658475, 2.2649786012211583, -0.35569683854275935, -1.4359238902281175, 3.545077379036972e-06, 1.536909083241135e-05, 0.1558874727010159, 0.1490695645278477, -1.476215756831183e-06, -1.6714741541188154e-05, -0.2841868412861729, 0.2319020915422329, 0.19878522901753773, 1.1602598807130757, 0.5697092793907033, 1.6460801018152058, 1.0710352577599354, 0.03824250540703592], "ratio96": 0.9999999999695982, "ratio_final": 0.9999999999783298, "diag": {"ratio": 0.9999999999783298, "nonnormal": 0.021012667300176053, "eig_gaps": [0.17338876608481182, 0.8833327990977008, 0.9947101204093693], "discness": 0.49701356770223437, "peaks": 1, "deg": 3, "diamW": 4.190209405764348}, "class": "one-exact", "probe_best": 0.9999999999796555} +{"family": "companion", "seed": 504010, "x": [-0.8215443934040554, -0.6228458326789117, -0.33374409125174015, 0.3217756429767192, -0.4143272225473712, 0.9807316949213511, 1.0174919761974088, -0.21145016807395395, 0.44900535618942333, 0.0931431218542505, 0.903129876288792, -0.05311013154509401, -0.1124869797065793, 1.0512284611332907, -0.13131284932008014, 0.8738977090613216, 0.062071235256496965, 0.3277672309813259, 0.9293311315118112, -0.1939057579104183], "ratio96": 1.781569880435885, "ratio_final": 1.7811672748407623, "diag": {"ratio": 1.7811672748407623, "nonnormal": 0.6471454576253582, "eig_gaps": [0.30106267093954775, 0.48213216759978805, 0.7503040513085761], "discness": 0.3203597549444552, "peaks": 4, "deg": 3, "diamW": 2.205072398324683}, "class": "below2-max", "probe_best": 1.781667098094651} +{"family": "peaks", "seed": 505010, "x": [-0.8938077191292364, -0.03995844874084214, -0.09299238203621657, -0.24614483778035257, 0.8634686811648897, -0.7157701802341019, 1.863685041112002, -0.11689580949523709, -0.054130436146221844, -0.00998765436725401, 1.198968490747382, -0.7042189561422916, -1.7950439154054303, 0.1920939979283788, -1.0765951830817087, 2.3223057305486856, 2.607715939241275, 1.2370506358978002, 0.26659181747475125, 0.6184857474955132], "ratio96": 1.9391649252978824, "ratio_final": 1.9390347714661518, "diag": {"ratio": 1.9390347714661518, "nonnormal": 0.6973256228617007, "eig_gaps": [0.27721233700498793, 0.35719894722335194, 0.6311529056731684], "discness": 0.11413065497735517, "peaks": 4, "deg": 3, "diamW": 2.9830306796896435}, "class": "stall", "probe_best": 1.9515239937318438} +{"family": "real", "seed": 506010, "x": [-0.30300505942341416, 0.05581254866580168, -0.4681059222307944, 0.0910934244737035, -0.31519384573231235, 0.07420613671943199, 0.10151333879643479, 0.224615332882677, -0.16392417831146894, -0.18639897430641034, 0.19978906018473996, -0.14112905485341176, -1.038891334716478, 0.3532666414999521, -1.4595864301067232, 0.13208684661047637, 1.4757366877989033, 0.4105837599662385, -1.5170092566579978, 0.8341108793759731], "ratio96": 1.9238737080722101, "ratio_final": 1.92459646692251, "diag": {"ratio": 1.92459646692251, "nonnormal": 0.43415553723606426, "eig_gaps": [0.04971525599146524, 0.3466156724318241, 0.38038163585860635], "discness": 3166252771.601639, "peaks": 3, "deg": 3, "diamW": 0.44383956319020795}, "class": "stall", "probe_best": 1.9251244235598899} +{"family": "random", "seed": 500011, "x": [-1.473690819064787, -1.0990063480932855, -1.004065792241105, -0.7595595406262587, -1.6786200934563809, -0.9957916245033859, 0.6245556908987903, 0.6758564725816174, -0.6040257987983245, 0.08820986588481011, -0.32249841212297925, 0.4587070649573013, 2.09788538003055, -2.3858020390617813, 0.16941993573760072, 1.4098915687896403, 1.8854594944968994, 2.4497243830690696, 0.569310037007658, 0.17011387788199195], "ratio96": 1.946229689354994, "ratio_final": 1.9468717506582869, "diag": {"ratio": 1.9468717506582869, "nonnormal": 0.22725309765813118, "eig_gaps": [0.166230490628938, 0.4197943131528951, 0.5177883286335802], "discness": 27355515619.89297, "peaks": 4, "deg": 3, "diamW": 1.3803383653629968}, "class": "stall", "probe_best": 1.9536228913531077} +{"family": "jordan3", "seed": 501011, "x": [0.2679801088664263, -0.6576042731060547, -0.4017001054304554, 0.4159664871473412, 0.05143550202829378, -0.09757391741934052, 0.7720539433018223, -0.568303051871085, 0.6977295984860811, -0.18546908660306427, 0.9691155709473218, 0.47505759888485, -0.02164931487862244, 0.5265925260153557, -0.18742609015410508, 0.32657545019513434, 1.6722828930719043, 0.4789659172663262, -0.6492078999707998, 0.271807529416459], "ratio96": 1.8676339146531065, "ratio_final": 1.8942031515403743, "diag": {"ratio": 1.8942031515403743, "nonnormal": 0.8486313441065582, "eig_gaps": [0.2927529933277705, 0.33392234781613395, 0.6169251443189715], "discness": 0.10819810592332903, "peaks": 4, "deg": 3, "diamW": 2.0510048595678487}, "class": "stall", "probe_best": 1.9521275872677664} +{"family": "jordan2", "seed": 502011, "x": [-0.07291649359855701, 0.05069778413274992, -0.026673164111510023, -0.41272516733249964, 0.14242686631564067, -0.6632184216347453, 1.3312381233144934, -0.06973118257026525, -0.3212145785011361, 0.007034330944552764, 0.06460651799977152, -0.09537007915080364, 0.15281159844019687, 0.24566549209876798, 1.454700660179582, -0.5893191638429898, 0.10220907457652678, 0.015302873767812905, 0.07209428780719035, -0.042086791474647486], "ratio96": 1.9919452121747758, "ratio_final": 1.9919396766211583, "diag": {"ratio": 1.9919396766211583, "nonnormal": 1.109452243204337, "eig_gaps": [0.2086469093707921, 0.32151888387871025, 0.5147946306164923], "discness": 0.024791324642718007, "peaks": 4, "deg": 3, "diamW": 1.4485135664854916}, "class": "near2-other", "probe_best": 1.9919396766211583} +{"family": "nearnormal", "seed": 503011, "x": [-0.9205846446704189, 0.998917286792752, 1.397950081278148, -2.5094662938852275, -0.6113718031148908, 0.5920468760546777, 1.2989118299948356e-05, -3.7256240554611837e-06, 0.09250258605115178, -0.07535377007866059, -7.037442821120625e-06, 1.9588421811929077e-05, -0.19071582508078275, 0.4151200316528447, 0.9138814384289267, 0.9286496460631006, 1.1973929689311482, 0.5816191581474008, 1.0506466144656166, 0.07789743267902545], "ratio96": 0.9999999999671868, "ratio_final": 0.9999999999789269, "diag": {"ratio": 0.9999999999789269, "nonnormal": 0.008171801821848188, "eig_gaps": [0.12132349232073236, 0.8773413898909357, 0.9983652638830948], "discness": 1.1436714048364587, "peaks": 1, "deg": 3, "diamW": 4.212163248354216}, "class": "one-exact", "probe_best": 0.999999999979713} +{"family": "companion", "seed": 504011, "x": [-0.5716366664945416, 0.6045527911611088, -0.41061333677342005, -1.8387077529433693, -1.2411734616367092, -1.6539199915564444, 1.1429305367984999, 0.2781900681762446, 0.3666930725918948, -0.29168350564045553, 0.9188761207951566, -0.04609510250764327, -0.07274282389224658, 0.03254605650524014, -0.316418392917439, -0.41177716411071685, 0.18581441598036957, 1.1603648858595639, 0.6421229654846909, 0.038335633365386146], "ratio96": 1.5387860370064235, "ratio_final": 1.5434588388660222, "diag": {"ratio": 1.5434588388660222, "nonnormal": 0.4324817373899727, "eig_gaps": [0.27923639213778334, 0.7730654712195839, 0.8035643414833571], "discness": 1.0738840594991081, "peaks": 4, "deg": 3, "diamW": 3.0471249129756526}, "class": "stall", "probe_best": 1.5516948806446815} +{"family": "peaks", "seed": 505011, "x": [0.21813704121953295, -0.7819493841535656, 0.5154947384272697, 0.36343674656551894, 0.8768566979396772, -0.2362177304777452, 1.3958065019404873, 0.24160682135849756, -0.05785612244237755, 0.22035610220728366, 1.637425540137683, -0.5070337100503619, -0.9033672067614452, 1.4385094420822824, 3.8781445013585962, -2.1611347557309393, -3.562468405709703, 0.6829253551022583, 0.14124264771409764, -0.8029538651155212], "ratio96": 1.960840831472854, "ratio_final": 1.9608644336249832, "diag": {"ratio": 1.9608644336249832, "nonnormal": 0.6438351270590273, "eig_gaps": [0.2779337564457444, 0.3395827655729416, 0.46976882013238064], "discness": 0.05865335756338977, "peaks": 3, "deg": 3, "diamW": 2.5190173673163088}, "class": "stall", "probe_best": 1.9796467730866658} +{"family": "random", "seed": 600000, "x": [-0.483786559547726, 0.529573307564668, 0.5423554662736787, 0.700604448364845, 0.2003892865476582, -0.9379511441330397, 1.166475726748983, 2.558035428850487, -0.7440065067172372, 0.38525720604892466, 0.37862213557250685, -2.4026219241308846, -0.5123909882802855, -0.7327540619511552, -0.053774589053468186, 0.7516250019247344, -0.7494083019999997, -1.8625815128691303, 0.0386318040013974, -0.23974496576024928], "ratio96": 1.9818846153263343, "ratio_final": 1.981859301889913, "diag": {"ratio": 1.981859301889913, "nonnormal": 0.7459930417423575, "eig_gaps": [0.24735364397783494, 0.3849942548879538, 0.3979969004630632], "discness": 0.027865492762448325, "peaks": 4, "deg": 3, "diamW": 4.205709642683019}, "class": "near2-other", "probe_best": 1.981859301889913} +{"family": "jordan3", "seed": 601000, "x": [0.17013622931512906, 0.220448812761221, 0.005114668177192909, -0.33876147527108125, 0.33224063773806956, 0.2614122286241456, 1.249578181882702, -0.4932143040228865, -0.08949146000037389, -0.2623022628289994, 1.2549500045344388, -0.38484419248921137, 0.04871542618851671, -0.13278253016951969, -0.6369782854536392, 0.4215742543102404, 1.2511076363124454, -1.538119289897737, -0.06091589287167187, -0.18932046413297807], "ratio96": 1.9894734611921723, "ratio_final": 1.989449813565899, "diag": {"ratio": 1.989449813565899, "nonnormal": 0.6984908041111284, "eig_gaps": [0.08228814466136145, 0.2869507767408545, 0.33640438373228354], "discness": 0.02742435843436653, "peaks": 3, "deg": 3, "diamW": 2.031884455404855}, "class": "near2-double", "probe_best": 1.989449813565899} +{"family": "jordan2", "seed": 602000, "x": [0.40821028860113295, 0.026852762456371027, 0.6022418286059905, 0.12329976786161212, 0.5813308957274721, -0.22881317559643946, 0.9770874542161818, 0.12097579390298441, -0.10529858427872957, -0.1492703765558006, 0.08298078652475205, -0.0667404997573717, -1.1716190168587466, -0.37069708386950606, 1.971816620765022, 0.3335332259326218, 0.48543589800715004, 0.029920339133090806, -0.07386335418940909, 0.10318390672625373], "ratio96": 1.9818427901872062, "ratio_final": 1.9825064773241765, "diag": {"ratio": 1.9825064773241765, "nonnormal": 0.7475589862246607, "eig_gaps": [0.2104080463602655, 0.29982743730084205, 0.3425230546408032], "discness": 4751874114.8994, "peaks": 3, "deg": 3, "diamW": 1.029808974916182}, "class": "near2-other", "probe_best": 1.9825064773241765} +{"family": "nearnormal", "seed": 603000, "x": [0.17560300293612, -0.8758908945058206, -0.8846560287947032, 0.2790137031652996, 0.6006684138364444, 0.13416260666577498, -1.3807821069054902, -0.6539790050433364, -0.283450992314791, -0.45539637026866453, 1.3227907227442048, 1.1125207926023055, -0.4033130964607201, 0.27507187561047325, -0.8483384009377757, 0.6257194812591425, 1.9458014465347626, 2.328222719650556, -0.3924442337197076, 0.756101231141817], "ratio96": 1.7855173202042904, "ratio_final": 1.7996378111063278, "diag": {"ratio": 1.7996378111063278, "nonnormal": 0.753456733493225, "eig_gaps": [0.38361276866868, 0.5224183410484214, 0.5488182450762725], "discness": 0.04080561542980108, "peaks": 3, "deg": 3, "diamW": 2.8566584690387633}, "class": "stall", "probe_best": 1.8034473914685982} +{"family": "companion", "seed": 604000, "x": [-0.061057044337958555, 0.10831093193712528, 0.7161448110956626, -0.6433090132762234, 0.1294673342888505, -0.3804555954345675, 1.7385530025948759, -0.5536289050387474, -0.7379542789233176, 1.4368903197948648, -0.1268014088762755, -0.46160087865000216, 0.7756087761415511, -0.3538160998075639, -2.4134113395612022, -1.004226619284542, -0.09405303233132878, -0.2670455903110769, 0.046233547312059683, 0.08527236969806083], "ratio96": 1.8557367583138789, "ratio_final": 1.8609734670980096, "diag": {"ratio": 1.8609734670980096, "nonnormal": 1.1679040316395943, "eig_gaps": [0.1987212374073686, 0.24352840430749703, 0.4095706607167262], "discness": 0.02696145036959189, "peaks": 3, "deg": 3, "diamW": 2.639817006388698}, "class": "stall", "probe_best": 1.86603412445619} +{"family": "peaks", "seed": 605000, "x": [-1.3992271248326735, -0.4012883207271959, 0.49752713679492966, 0.29913697400388584, 0.8168948628427053, 0.7792848628386229, -0.0211200031532976, -0.020631587334709645, 0.012599224684066507, 0.0003119323865111042, 1.3627112661286647, 0.5176860258374449, 0.05773970963496267, 2.4983360815210283, -1.4228132366831063, -0.9022769802795925, -1.522020238865631, -1.259677459075421, 0.47320663672620067, -0.1749676393562627], "ratio96": 1.8361126677437953, "ratio_final": 1.8222197897543364, "diag": {"ratio": 1.8222197897543364, "nonnormal": 0.5519545844368837, "eig_gaps": [0.1902583286280776, 0.667103340601288, 0.8284456471017063], "discness": 0.35975411113759553, "peaks": 1, "deg": 3, "diamW": 3.030936187345354}, "class": "stall", "probe_best": 1.8520971568259785} +{"family": "real", "seed": 606000, "x": [1.45320120177787, -0.013891352829364947, 2.0296394290376654, -0.002277076993149135, -0.4034640594559853, 0.01035145352354874, 0.9610140301140722, -0.5611309701090141, -0.697575870822283, -0.1963260956231899, 0.8226019228077479, 0.6690753570423664, 0.2045267341945219, 0.019920258877799853, -1.2861421866045128, 0.08350938947337666, 0.42479634958889745, -0.0001747953236459178, 0.11275950813625019, -0.03369594833746278], "ratio96": 1.8205397543588053, "ratio_final": 1.8214054701696325, "diag": {"ratio": 1.8214054701696325, "nonnormal": 0.4290727296076895, "eig_gaps": [0.19092983076300446, 0.6148986072508711, 0.8057481378677556], "discness": 0.7679564489364819, "peaks": 4, "deg": 3, "diamW": 3.0197230956385126}, "class": "stall", "probe_best": 1.9182604150686156} +{"family": "random", "seed": 600001, "x": [-1.3033134754288342, 3.473470684357058, -1.1244929774053012, 3.3355258263837033, -1.3724416716476298, 3.492839620071009, -0.16924618424847737, -0.7048133450411078, -0.1604740362582054, -0.6136749558903021, 0.4292459602290378, -0.08729405457309519, 0.3539675105223247, 0.7601813939566148, 6.502550729283662, -1.4574504121284626, -1.0934595221238421, -4.530156447007874, 1.3613601012332595, -1.0508601932470936], "ratio96": 1.8549802017412118, "ratio_final": 1.854996819679032, "diag": {"ratio": 1.854996819679032, "nonnormal": 0.0372557468219564, "eig_gaps": [0.06677194676607481, 0.21005660006871799, 0.27311602356354747], "discness": 8320025814.095204, "peaks": 3, "deg": 3, "diamW": 1.0751582809203684}, "class": "stall", "probe_best": 1.8929170885625728} +{"family": "jordan3", "seed": 601001, "x": [-0.10363451909761393, -0.27102920609892656, -0.1637355825105048, 0.6696457385389887, 0.7242292325480246, 0.21745793196418423, 1.6906500622286114, 0.6547665917668324, 0.018607580907323645, -0.031165020490362655, 1.5785889810096232, 0.8475985950010101, -0.02245939588893113, -0.0333091468646395, -0.17065767627852296, -0.2140539561843916, 0.5294367819331034, 0.016316177698056322, 0.045615336125179484, 0.08893786673681311], "ratio96": 1.9745921289102404, "ratio_final": 1.9746157296609834, "diag": {"ratio": 1.9746157296609834, "nonnormal": 0.7657367809859611, "eig_gaps": [0.34540419937906003, 0.3522364166937383, 0.3651474532270104], "discness": 0.012803501749542945, "peaks": 4, "deg": 3, "diamW": 2.7289562844117268}, "class": "near2-other", "probe_best": 1.9746157296609834} +{"family": "jordan2", "seed": 602001, "x": [0.4703733981271889, 0.17602323463017222, -0.16319766013510872, 0.19558646468505225, -0.1740832501728684, -0.046147379095912704, 1.25971805957531, 0.6782201087559052, -0.1107482966304674, 0.013433261799948387, -0.021384370345143906, -0.012988988565902766, -0.201842161722655, -0.17973577167366867, 1.1586349583294293, -0.08563722617628447, 0.06881238065147963, 0.0903462963738606, -0.09114138422143683, -0.0020345386195672333], "ratio96": 1.984552682755901, "ratio_final": 1.9845199763927919, "diag": {"ratio": 1.9845199763927919, "nonnormal": 1.3212748496762168, "eig_gaps": [0.15424612227294726, 0.40405378076271703, 0.4345260711305998], "discness": 0.0463151630391169, "peaks": 4, "deg": 3, "diamW": 1.5687837868577623}, "class": "near2-other", "probe_best": 1.9845199763927919} +{"family": "nearnormal", "seed": 603001, "x": [1.5171417610202846, -0.314308209807026, 2.1592002016991674, -0.8983639259723446, 0.4378371015782165, -1.1703854139437566, 1.4453998406491662, -0.9033338868668751, -0.13938242631533693, 0.06915962740152853, -0.27556558469150216, 0.03221817689522289, 5.607670186294116, 5.803184301685721, 3.5824484174565607, -7.77883454196755, -4.808512632376088, 1.3028887180205353, 0.6749335884664274, 0.20362779175707438], "ratio96": 1.7413819131845125, "ratio_final": 1.7640525690219728, "diag": {"ratio": 1.7640525690219728, "nonnormal": 0.3802811793069233, "eig_gaps": [0.36049528520543034, 0.5721628942153648, 0.7238138088419271], "discness": 122422490027.4814, "peaks": 3, "deg": 3, "diamW": 2.4076964057098773}, "class": "stall", "probe_best": 1.813616832319098} +{"family": "companion", "seed": 604001, "x": [-2.086126728161955, -1.8662579453717205, -0.06059696334083362, -2.5530890453583446, 2.0736391020752825, -0.644460916214421, 0.986738812711021, -1.1250644507622223, -0.6313582118774516, -0.2143536143034987, 1.10573696145924, -0.9608745786595287, 0.4594121758809362, 0.29620184571951286, 2.5833749182526535, -0.39563570663360914, 0.4942183466581358, 0.26233876160417813, 1.0186721146891733, 0.5644404239667464], "ratio96": 1.380750452255246, "ratio_final": 1.3853016218976717, "diag": {"ratio": 1.3853016218976717, "nonnormal": 0.36380431435456173, "eig_gaps": [0.4392573058339129, 0.5880254558585318, 0.8903988824862764], "discness": 1.1809878114992287, "peaks": 3, "deg": 3, "diamW": 4.869150430817396}, "class": "stall", "probe_best": 1.391014470475884} +{"family": "peaks", "seed": 605001, "x": [0.5075523671628827, -0.04077685050903449, 0.2739199379074265, -0.2547681366179897, 0.8869633908345655, -0.15584063867745312, 0.020000406074527438, -0.003521511219083296, -0.027049799649832798, 0.030892482765432196, 1.1489495227561095, -0.02070481291939345, 1.8714471069203182, -0.5194614532265609, -2.723992102676408, -0.5912250692226877, -0.9999774205534904, 0.67693197868297, 0.5297435914055975, -0.18295322875465747], "ratio96": 1.9699480337148254, "ratio_final": 1.9704061233923598, "diag": {"ratio": 1.9704061233923598, "nonnormal": 0.8391764379928809, "eig_gaps": [0.24240208713221226, 0.303344662379142, 0.47510994749862395], "discness": 27740797276.22095, "peaks": 4, "deg": 3, "diamW": 1.3070115190623237}, "class": "near2-other", "probe_best": 1.9704061233923598} +{"family": "real", "seed": 606001, "x": [-0.3815346818587016, -0.162211618284429, -0.8334672516783328, 0.04268669202973023, 0.30425181912458943, -0.07419620983558725, -0.3495436053102085, -0.1831020152584926, 0.570722526267198, -2.032522804166162, -2.730300068476479, -0.7409386328794633, 0.7465685530222768, -0.24806636396525467, 4.23565055875034, 0.1833471801034355, 0.05715404116321318, -0.045862820702841596, -0.03090234514853768, -0.012043370145907787], "ratio96": 1.990399538287113, "ratio_final": 1.990434852368694, "diag": {"ratio": 1.990434852368694, "nonnormal": 1.3688252985267515, "eig_gaps": [0.13310747515979288, 0.18546918168868187, 0.3067962497457659], "discness": 0.021204701097621424, "peaks": 4, "deg": 3, "diamW": 3.7279049566377154}, "class": "near2-double", "probe_best": 1.990434852368694} +{"family": "random", "seed": 600002, "x": [1.6168180598408255, 1.0304230373511154, 1.1258948508526974, -1.1176309985805368, 0.8694173032349636, 1.6086483497604227, 0.11809189896994073, 1.7782359744623315, -0.9447811754758331, 0.7814795212288858, -1.7586295196440793, -0.5221822126698483, -2.5662433710206507, -2.880921400778191, 1.1299966145080003, 2.2855052766806985, -0.16188116958400536, 0.6604806883484793, -0.9213061301168428, -0.5488532631937661], "ratio96": 1.5326847834451718, "ratio_final": 1.5989092253951127, "diag": {"ratio": 1.5989092253951127, "nonnormal": 0.5815032511067285, "eig_gaps": [0.2353978807004419, 0.5488950950282966, 0.682137791760927], "discness": 0.3209466746461186, "peaks": 3, "deg": 3, "diamW": 4.014316463835713}, "class": "stall", "probe_best": 1.6014423707757572} +{"family": "jordan3", "seed": 601002, "x": [0.4910859889433299, -0.5406588116263269, -0.14370609615466085, 0.004371559501834217, -0.5634969810799144, 0.23181768057059615, 1.1650093050227408, -0.5722097351445888, 0.21033693109984034, 0.04829068608793158, 1.2615093689536807, 0.020630260335673868, -0.0024845885977623624, 0.3409444791336388, 0.33275778132387335, 0.23588267067740248, 1.414047809357368, -0.40207553019282566, -0.1459259817764797, 0.02366637231059926], "ratio96": 1.9703536012156118, "ratio_final": 1.970341072659899, "diag": {"ratio": 1.970341072659899, "nonnormal": 0.6951641486102499, "eig_gaps": [0.21237452306232957, 0.3721617434121056, 0.581474511962557], "discness": 0.08370423617935056, "peaks": 4, "deg": 3, "diamW": 2.2481392936291766}, "class": "near2-other", "probe_best": 1.970341072659899} +{"family": "jordan2", "seed": 602002, "x": [-0.08855164460637693, -0.19335783108193771, 0.44139881227160294, 0.2087842991615908, 0.4580486509190249, -0.18558394670997136, 1.543179191826876, 0.4915964812787917, 1.3270035046667061, 0.4101281114578845, 0.05353399466636341, 0.1624163897077251, -0.2795943186190175, 0.1423045742338035, 1.329448510615379, -0.31451427212993044, 0.032566710083742303, 0.002083974443680742, -0.0017838683005976354, 0.02075566429182786], "ratio96": 1.9930090616710385, "ratio_final": 1.9930091694590535, "diag": {"ratio": 1.9930091694590535, "nonnormal": 1.3333518076396678, "eig_gaps": [0.17628378837349462, 0.24413919507138884, 0.29710683491840184], "discness": 0.01505245952878435, "peaks": 3, "deg": 3, "diamW": 2.2391143446750967}, "class": "near2-other", "probe_best": 1.9930091694590535} +{"family": "nearnormal", "seed": 603002, "x": [0.6601748067002443, -0.3852464793174133, 0.5418962584423367, -0.24186584921444648, 0.691228780783226, -0.36819393026470926, 0.23708897171391075, -0.1514582651326482, 0.4597487272978425, 0.005316828414408967, -0.11581094322081892, 0.10750526541881955, -0.3123960418724254, -0.7938539187159114, 0.6070299079980277, 0.4246699897177282, -1.8452587320587757, 0.4969921335825642, 0.6989074691079303, 0.2861574784839396], "ratio96": 1.9693415847322253, "ratio_final": 1.9718211928815372, "diag": {"ratio": 1.9718211928815372, "nonnormal": 0.252494782926166, "eig_gaps": [0.058742958064697266, 0.3081912258130484, 0.32432219832518666], "discness": 7608349979.139288, "peaks": 4, "deg": 3, "diamW": 0.6031010079411856}, "class": "near2-double", "probe_best": 1.9718211928815372} +{"family": "companion", "seed": 604002, "x": [-0.5720987789816994, 0.21343716876010665, 2.174800486950332, 0.015844027870660568, 0.08288124358921045, -0.1666925739451262, 1.4745224360668778, -1.6762965623304682, 0.9562328585367572, -1.6511339225553354, -2.014858773720798, 0.9298392462524004, -0.6605585458139289, 2.4221014130959535, 0.024518344538230777, 5.809574643793576, -0.3355020491410424, -3.868984646818368, 0.30800353932527325, -0.3440349168978257], "ratio96": 1.960300814195675, "ratio_final": 1.96025972749273, "diag": {"ratio": 1.96025972749273, "nonnormal": 0.6875588886467539, "eig_gaps": [0.1625469072829485, 0.4507180655461723, 0.5911210203469861], "discness": 0.10650156123378811, "peaks": 4, "deg": 3, "diamW": 4.658939119018658}, "class": "below2-max", "probe_best": 1.960499365306293} +{"family": "peaks", "seed": 605002, "x": [-1.4925882664586005, 0.22958050930881915, -0.30763699049448157, 0.24029363769638973, 1.143469430973672, 0.09776819068140394, 1.8646257084767965, -1.1146877796814072, 0.43959294982178226, 0.1255937373786603, 2.572862822588974, 0.5829737647250971, 1.6906166911383063, 0.11922130696511529, -2.068431648577338, 1.1635896619729886, -2.635451599414301, 0.232987259686058, 0.38348271450930105, -0.0027854281294325306], "ratio96": 1.9684686596695675, "ratio_final": 1.9702663383359411, "diag": {"ratio": 1.9702663383359411, "nonnormal": 0.7238337601804495, "eig_gaps": [0.27452432021402057, 0.33778985188400573, 0.6114483297708638], "discness": 0.0860870436540157, "peaks": 4, "deg": 3, "diamW": 4.316556371671547}, "class": "near2-other", "probe_best": 1.9702663383359411} +{"family": "real", "seed": 606002, "x": [0.1195065285189599, 0.21281077765666173, 0.27671792554689356, -0.024608702079277435, -0.3185470749662682, -0.05573117268874665, 0.3645802000314601, 0.08233915712455124, 2.2046211739807435, 0.5879001027687792, -0.5055032989779731, 0.17670315091212666, -0.4510536339405876, 0.2236613057483524, -2.2946198968908362, 0.49490133216288523, 0.10487045411951756, -0.09460508442749552, -0.005665638640628461, -0.012697009269545305], "ratio96": 1.9858874406934037, "ratio_final": 1.9879277920417806, "diag": {"ratio": 1.9879277920417806, "nonnormal": 1.370120407306207, "eig_gaps": [0.11759881303826007, 0.21219911025831287, 0.24617276119014245], "discness": 0.010812234155000792, "peaks": 3, "deg": 3, "diamW": 2.421380970351202}, "class": "near2-double", "probe_best": 1.9879277920417806} +{"family": "random", "seed": 600003, "x": [0.04790071821568607, -1.1346993324158836, 0.33685352483381237, -0.7878878945287856, 0.1069758807235365, -0.09010455946111502, 0.2582944118462696, -0.7152691015907005, -0.07417576466374047, 0.4301075111850337, 0.8764357779664764, -0.5177489961251561, -1.1343909380591433, -0.3830731987784674, -0.13370011236780474, 3.019055114930917, 0.4576524345897668, -0.33575467465521547, 0.0688903303364937, 1.3477796628455714], "ratio96": 1.872437307991583, "ratio_final": 1.8970449526308026, "diag": {"ratio": 1.8970449526308026, "nonnormal": 0.5110848592249776, "eig_gaps": [0.2660734015210353, 0.43303575500863006, 0.6166951750809605], "discness": 0.36871219312864045, "peaks": 4, "deg": 3, "diamW": 1.6965657038684547}, "class": "stall", "probe_best": 1.8979254551164828} +{"family": "jordan3", "seed": 601003, "x": [0.14409124557984498, 0.01290867607529127, -0.2476299908580556, -0.15265270808686954, -0.2656237136393142, 0.2971042870153203, 1.682474274192485, -0.001903370764457356, 0.31839147840455834, 0.09862497431122462, 1.3526341931595554, 0.9268646512143232, 0.035146139043216826, -0.06271585597806621, 0.29667123709006327, -0.39146340987576755, 1.2286730176438527, -0.8049429076784971, -0.1749605658234433, 0.04867772760466256], "ratio96": 1.9948218841283025, "ratio_final": 1.994816612552754, "diag": {"ratio": 1.994816612552754, "nonnormal": 0.7182665573214544, "eig_gaps": [0.17581791556602952, 0.1860894553524695, 0.2061466886692669], "discness": 0.0020367420389539167, "peaks": 3, "deg": 3, "diamW": 2.4188194525083984}, "class": "near2-other", "probe_best": 1.994816612552754} +{"family": "jordan2", "seed": 602003, "x": [0.5153699363202136, 0.19462389284054138, 0.3573371927568145, -0.15232418478818666, 0.3400632757232717, -0.2952172375032175, 0.8224801157559094, -0.32885844908902384, 0.24769124547840723, 0.41133132948321616, -0.16850143448364185, 0.021352286743649404, -0.2221471973357515, -0.4887496287208082, 0.46751758606462124, 0.819849261297164, 0.32112337245978306, 0.07681885291822706, 0.0028888809732940854, -0.012660217472187148], "ratio96": 1.9262834262497317, "ratio_final": 1.9316633108665184, "diag": {"ratio": 1.9316633108665184, "nonnormal": 0.8993388479980872, "eig_gaps": [0.133636299306548, 0.35397001935959044, 0.4830459594618915], "discness": 0.11412274924062075, "peaks": 3, "deg": 3, "diamW": 1.0770528878878565}, "class": "stall", "probe_best": 1.932289853941436} +{"family": "nearnormal", "seed": 603003, "x": [-0.6924544199634555, 0.19307787059626738, -0.3228443835266245, -0.963790739303729, -0.21966786000434335, -1.0069693356746352, -0.028835706002084634, 0.01461277912570846, 0.004704060735565471, -0.04061763174301851, -0.2737481642795961, 0.173050301058706, -0.11230452190153656, 0.1831602076179899, 0.12890193790651955, 1.6050918792666096, 1.323595276341583, 1.7293634717967015, 1.2591644683868632, 0.34084754650970384], "ratio96": 1.719009451637423, "ratio_final": 1.7187594873165428, "diag": {"ratio": 1.7187594873165428, "nonnormal": 0.06719854514291566, "eig_gaps": [0.07884619484016314, 0.8561412258618379, 0.9092546446864818], "discness": 187528502750.3034, "peaks": 2, "deg": 3, "diamW": 1.418548551575718}, "class": "stall", "probe_best": 1.7534217320572154} +{"family": "companion", "seed": 604003, "x": [0.6323854176586391, 1.3007272389119027, -0.8901792033258537, -1.211673903641992, 1.3410040970172952, 0.06532475771371864, 6.313453073383169, -3.5666102411055536, -1.5666785905773482, -0.7487794628090111, 0.32391473487867206, 0.008852522752009907, -0.23664681590875725, 3.833933866035115, -6.048535569459153, 0.15038548532386964, -0.011532710474450314, 0.19569355005625533, 0.002141111404760797, -0.009348803527375237], "ratio96": 1.943821970391539, "ratio_final": 1.9462486629760212, "diag": {"ratio": 1.9462486629760212, "nonnormal": 1.3603102259622077, "eig_gaps": [0.1770560549096928, 0.31959724890644126, 0.36521853798192055], "discness": 0.027447398942766495, "peaks": 3, "deg": 3, "diamW": 8.043808188303133}, "class": "stall", "probe_best": 1.9477478867047309} +{"family": "peaks", "seed": 605003, "x": [-0.8927896417298795, -0.10143110142024458, -0.6643006988119424, 0.19513225065728135, -0.047450026984793475, -0.262837344669258, 1.1368341603378456, -0.12098354743315333, 0.4001210386490157, 0.1591658163989167, 1.2645873535874668, -0.18803253331106912, 0.8219215102112075, 1.2969157341758375, 4.066955999288061, 3.0148679887668344, 2.8880705650595426, 1.625775116436719, -1.203904006457277, -0.4385201991653698], "ratio96": 1.9482314751347063, "ratio_final": 1.9504934396249074, "diag": {"ratio": 1.9504934396249074, "nonnormal": 0.60640588469911, "eig_gaps": [0.18952161791740535, 0.3889251072734205, 0.43567049380474654], "discness": 0.06345218243591699, "peaks": 3, "deg": 3, "diamW": 1.9753709852752992}, "class": "stall", "probe_best": 1.9545489017049513} +{"family": "real", "seed": 606003, "x": [-0.2566913770005327, 0.03475889703430658, -0.4265388424122325, -0.029481609145788693, -0.2559070395181571, -0.01077550647922096, 0.1197575394817135, -0.12516671381894157, -0.1980977003290842, 0.15300252886338317, 0.29661007428315905, 0.19236383008160923, -1.609952808190609, 0.300473477120563, -2.6355086688001985, 0.7205942225181977, 2.5684039858436325, -0.7440482499639566, -1.7943581229892072, 0.39813620364895624], "ratio96": 1.9284490194505288, "ratio_final": 1.9289403584145925, "diag": {"ratio": 1.9289403584145925, "nonnormal": 0.601821963989707, "eig_gaps": [0.09286620456983961, 0.3500320480904181, 0.3702934708886298], "discness": 4365565073.440474, "peaks": 3, "deg": 3, "diamW": 0.49039538542162986}, "class": "stall", "probe_best": 1.929475352594001} +{"family": "random", "seed": 600004, "x": [0.2731449129121549, -0.3500702624419034, -0.6573614232027121, 0.8320932219699912, -0.16246777743518864, 1.3657054533125472, -0.44282512979478544, -1.7800595586247796, -0.7200570649885343, 0.3713423840075123, -1.039382051891872, -0.8644479389560134, 0.108941250669809, -0.3425742722261296, 1.5211945338648252, -0.27699368388468393, 0.2898597818059422, 0.7415428800785169, 0.1424852130194656, -0.2613437681483193], "ratio96": 1.9347212377287883, "ratio_final": 1.9349931508332516, "diag": {"ratio": 1.9349931508332516, "nonnormal": 0.7023757408064905, "eig_gaps": [0.24401368949716779, 0.5044184874420107, 0.5935260307176465], "discness": 0.09189893234885027, "peaks": 4, "deg": 3, "diamW": 2.9825319691898393}, "class": "stall", "probe_best": 1.9527661750171843} +{"family": "jordan3", "seed": 601004, "x": [0.30097624973635306, -0.05041173849327954, 0.10018271146439672, 0.04495547149506737, 0.08506834075734643, 0.2170800001336809, 1.6086043338609135, 0.2591359940466885, -0.12448829423257406, 0.12820382686810083, 1.6123930833458768, -0.014167380683331223, 0.003901834878486419, 0.09672017407290046, -0.38205719883502653, -0.40954762626773467, 1.3200504852509873, 0.6236482879770664, 0.15273029596051396, -0.04462918478146716], "ratio96": 1.9952131413435652, "ratio_final": 1.995203391499658, "diag": {"ratio": 1.995203391499658, "nonnormal": 0.716538156199698, "eig_gaps": [0.0743162227720351, 0.09560779804640068, 0.14785055878771358], "discness": 0.005559526964204263, "peaks": 3, "deg": 3, "diamW": 2.325022039919355}, "class": "near2-triple", "probe_best": 1.995203391499658} +{"family": "jordan2", "seed": 602004, "x": [0.2053750906049205, 0.0984773580689698, 0.3270816828526707, 0.1713600972420975, 0.3008895534780897, 0.27808200848351106, 1.184327514954918, 0.011832699294115635, -0.04092432970748106, 0.2660205160723768, 0.023738699977539492, 0.016751884942643288, -0.37629028595993685, -0.24458395328426563, 1.355185096915448, 0.19196923022714854, 0.1017457866643933, 0.014654339668854847, -0.0020744639488088287, 0.01324687498673336], "ratio96": 1.996692475989796, "ratio_final": 1.9966914813962127, "diag": {"ratio": 1.9966914813962127, "nonnormal": 1.1470378927922793, "eig_gaps": [0.08982265697436022, 0.11595593714703796, 0.16627666816166164], "discness": 0.004930178870139324, "peaks": 3, "deg": 3, "diamW": 1.2233996549532116}, "class": "near2-double", "probe_best": 1.9966914813962127} +{"family": "nearnormal", "seed": 603004, "x": [-1.233804600737956, 1.434234790212786, 1.7798159927375434, -0.7711993391473615, 0.4893245145874551, 2.195832340277118, -5.514007770573618e-06, 1.501263011935098e-05, -0.10906481812564148, -0.048332070135748334, -5.506079061766619e-06, 4.5498893965717527e-07, 0.3988553690165122, 0.7742252331025311, -3.044800532973633, 0.9142560411014835, -0.2521420665474901, -2.976458866236585, 0.8868684329476836, 0.11682891406881923], "ratio96": 0.9999999999708791, "ratio_final": 0.9999999999855295, "diag": {"ratio": 0.9999999999855295, "nonnormal": 0.025647633594428662, "eig_gaps": [0.5039699034569374, 0.8655343623809016, 0.9989909382798533], "discness": 0.7027323326576764, "peaks": 1, "deg": 3, "diamW": 3.7381861753118653}, "class": "one-exact", "probe_best": 0.9999999999864347} +{"family": "companion", "seed": 604004, "x": [-0.9783212171591003, -1.0891028062411447, -1.3415849241803706, 1.2836160748262473, 1.7669446918734495, -1.0866782798562489, 1.0295226493508403, 1.7801751493193585, 0.644908815512653, 1.497295727925965, 1.5589510990177662, -0.6777479045453649, 0.23514093605357234, -0.22775789633578267, -0.18238405912874045, 0.06685600898848872, -3.4476573027790556, -0.29796630678873653, 0.5008223468861122, -0.21563136018131468], "ratio96": 1.5050741334613045, "ratio_final": 1.5758094914744551, "diag": {"ratio": 1.5758094914744551, "nonnormal": 0.6669731133671148, "eig_gaps": [0.47992008331853997, 0.5488783510163235, 0.781575576799967], "discness": 0.13101225153041857, "peaks": 3, "deg": 3, "diamW": 5.00159456932328}, "class": "stall", "probe_best": 1.578583715916359} +{"family": "peaks", "seed": 605004, "x": [-1.1278022501950402, -0.2545004242985418, -0.49831145644052044, 0.3959312904274337, -0.21415244657087218, -0.7684961500979393, 2.2490660018669537, 0.8247630737866214, -0.2873812977259188, -0.05858476607105431, 1.7122072006442177, 1.5531431266584623, 1.5293132299836891, 1.5931017502737865, 5.728348122426677, 1.7413248989161576, 4.339422109183824, -0.10410672036863039, 0.12597444124508556, 0.42926008312830444], "ratio96": 1.9785357609598182, "ratio_final": 1.978559453882369, "diag": {"ratio": 1.978559453882369, "nonnormal": 0.659681817852658, "eig_gaps": [0.2552931595531666, 0.29566564656698296, 0.33805393508673287], "discness": 0.01595833756472086, "peaks": 3, "deg": 3, "diamW": 3.545582700820795}, "class": "near2-other", "probe_best": 1.978559453882369} +{"family": "real", "seed": 606004, "x": [2.008530672267221, 0.7001503153799697, -1.2625604631327683, -0.7273530040644184, -0.4718457718448178, -1.1889037151020156, 0.38049367060081307, 0.6650160971787192, -0.7953869953075838, 0.4483143658894968, -0.013659823893765834, 1.0366196139922388, 1.2699382788615474, -1.0835473068251718, -0.29482541003532187, 0.1949345750478072, 0.7899334790728734, 0.8522037278132089, -0.14847195757798076, -0.23731752292111163], "ratio96": 1.5330881839246997, "ratio_final": 1.5463886315837072, "diag": {"ratio": 1.5463886315837072, "nonnormal": 0.5067020335211357, "eig_gaps": [0.22788154623029994, 0.7760164661060669, 0.8883157547038528], "discness": 0.37033360603735577, "peaks": 3, "deg": 3, "diamW": 4.01772214117389}, "class": "stall", "probe_best": 1.5469775938135835} +{"family": "random", "seed": 600005, "x": [0.48028691300790793, 0.29739542156791327, -0.802802739046464, 0.4327438149368178, 0.8262234040599581, -0.9219955483887792, 1.6402557552396264, -0.005957361847453296, 0.5817625276278966, -1.1999467837332773, -0.43176083211167154, -1.8518951111512871, 1.5709262823747234, 0.6918519743210033, -0.22568516899566776, 0.6113671666541207, 0.2225010720173538, -5.493720402867936, -0.7142547389157953, 0.5107630574576634], "ratio96": 1.9224561116401375, "ratio_final": 1.93390661994052, "diag": {"ratio": 1.93390661994052, "nonnormal": 0.7560016273359537, "eig_gaps": [0.36011756482072305, 0.3665660133607969, 0.6019622238356217], "discness": 0.07718849185434486, "peaks": 4, "deg": 3, "diamW": 3.519716933229399}, "class": "stall", "probe_best": 1.9362707163176809} +{"family": "jordan3", "seed": 601005, "x": [-0.21581488856420505, -0.2312006366628763, 0.3342442458054502, 0.12860888042082425, -0.15295109200542412, 0.43598757321482934, 1.0613697873143382, -0.6585645854574628, -0.48259362952924056, 0.23313805488615583, 0.9808945584988993, 0.537445429814906, -0.026119201572154582, 0.04689712031222895, -0.11154857086891568, -0.28327840866440523, 0.8182252098374758, -0.36801308920690623, 0.04634994813066974, -0.2757236319412787], "ratio96": 1.9797224204475175, "ratio_final": 1.9799329149467628, "diag": {"ratio": 1.9799329149467628, "nonnormal": 0.7473096235186139, "eig_gaps": [0.30740362785594216, 0.35075180242724496, 0.3576116318727301], "discness": 0.007350929189456908, "peaks": 4, "deg": 3, "diamW": 1.8739413987514382}, "class": "near2-other", "probe_best": 1.9799329149467628} +{"family": "jordan2", "seed": 602005, "x": [0.09003171710320756, -0.11999367405616038, -0.04972393654309234, 0.0649696553336624, -0.11304217890155073, -0.3102651707193421, 1.1701023271671742, -0.4319496726978295, -0.38033155215520764, 0.28180840616016983, -0.17372685736547455, -0.14435820549201978, -0.03396057984802256, -0.02368905940648666, 2.0018389465839466, -0.44668814322050787, -0.1255790252471154, -0.12956955082588978, -0.003755564314191627, -0.04302407488400947], "ratio96": 1.9848054989623232, "ratio_final": 1.9848139471528905, "diag": {"ratio": 1.9848139471528905, "nonnormal": 1.3115284842021386, "eig_gaps": [0.16864122859460587, 0.2024377390287048, 0.2768231053461325], "discness": 0.010463682004651162, "peaks": 3, "deg": 3, "diamW": 1.37466698409107}, "class": "near2-other", "probe_best": 1.9848139471528905} +{"family": "nearnormal", "seed": 603005, "x": [-0.4615300626789711, 1.0890592402493162, -0.18722824674137362, 0.10046490345884984, 1.0110325749667473, -0.498047744856593, 1.3319350599779447, 1.5527657059519153, 0.21324687129207498, 0.23547370873430024, -0.41246584466850045, -1.839371064046242, -0.8654448594187776, -0.43456796940202114, 0.6144858499739739, -0.051039055165771396, -0.5516101397678659, 0.9896464771867675, 0.013068680067292476, 0.18468721726382697], "ratio96": 1.9572253266066792, "ratio_final": 1.961538994240155, "diag": {"ratio": 1.961538994240155, "nonnormal": 0.7274247725660782, "eig_gaps": [0.2893917934557761, 0.3778153191033986, 0.610697846255246], "discness": 0.09148055244828178, "peaks": 4, "deg": 3, "diamW": 3.545171676464817}, "class": "stall", "probe_best": 1.9665905997661048} +{"family": "companion", "seed": 604005, "x": [-0.335436846829379, -1.7394397732873117, 0.7287195336855687, 1.7602315418217267, 2.9027513426756775, -1.131686202035291, 2.403590405312177, 1.0669907106789012, 0.8514686762105939, 0.09393232459041548, 0.23345850708455784, 0.17456354018869502, 0.9414431809590948, -0.8008476742602169, -4.428177584449417, 1.7434297409729322, -0.8419037108396537, -0.16476056474587775, 0.1991341105072167, 0.13307823508723526], "ratio96": 1.382959740674868, "ratio_final": 1.4635977531638367, "diag": {"ratio": 1.4635977531638367, "nonnormal": 0.7518629804715131, "eig_gaps": [0.7161557599392148, 0.7864140449756183, 0.7950934344633599], "discness": 0.13468952070588447, "peaks": 3, "deg": 3, "diamW": 4.600573448887192}, "class": "stall", "probe_best": 1.4646998315906716} +{"family": "peaks", "seed": 605005, "x": [-1.627479847110239, -0.44907584793851313, 0.36667760032044927, 0.0669178157957378, 1.1842336505033844, 0.30832263671788385, 0.6544949809742306, -0.24503301564341223, -0.12613521381345186, -0.025074256327523428, 1.8185328179291744, 0.08901639042234884, 1.5024878173197498, 2.189734413684656, -1.4398327683733747, -1.2118101475453684, -2.2159117714895036, -1.1100826253572693, 0.6169095746915572, 0.14696940247453338], "ratio96": 1.8269455289368668, "ratio_final": 1.8420652209399841, "diag": {"ratio": 1.8420652209399841, "nonnormal": 0.6384423064458211, "eig_gaps": [0.23874665202665238, 0.5768986302941969, 0.8155481530906135], "discness": 0.23314710751088577, "peaks": 5, "deg": 3, "diamW": 3.5705290864749113}, "class": "stall", "probe_best": 1.8447068883818174} +{"family": "real", "seed": 606005, "x": [0.23987281551422526, 0.3732522286416609, -0.11903510842712742, -0.23539821630134355, -0.19155447714502857, 1.2418454134945716, 0.04416553946799092, 0.16034823533391393, -0.9145589154594972, 0.1038836688359851, 2.9784996037081632, -0.31760326843155634, 0.18397015118884677, -2.709166571238137, 3.6511369665963773, -1.793958814283818, -0.31340318406686973, -0.06417194565381498, 0.07707734242861607, -0.046898270878417934], "ratio96": 1.9396567513890708, "ratio_final": 1.944038870107737, "diag": {"ratio": 1.944038870107737, "nonnormal": 1.3170615683170301, "eig_gaps": [0.20369750994558938, 0.27958674534164135, 0.4263759261860648], "discness": 0.051035558074515104, "peaks": 3, "deg": 3, "diamW": 3.468822914301325}, "class": "stall", "probe_best": 1.9690615551681572} +{"family": "random", "seed": 600006, "x": [-0.05095443983090295, -0.3607699703275107, -1.0622174445718864, 0.14920320973202159, -1.2184289258546386, 0.1904118445911489, 1.251366980958539, 0.08173938024498904, -0.41080683233957666, -0.7748136479309472, -0.10715713866567282, -0.0067463024039094086, 0.8635907270107408, 0.5829281140897684, 1.8254525307192297, 0.5265548675340221, 0.030096894333709018, -0.4106358024525291, -0.08529102670053346, -0.26148312494804715], "ratio96": 1.911470875578101, "ratio_final": 1.9153715426347542, "diag": {"ratio": 1.9153715426347542, "nonnormal": 0.8030130736600809, "eig_gaps": [0.08404877630480158, 0.5892187006007545, 0.6716625756644881], "discness": 0.2438831112923686, "peaks": 4, "deg": 3, "diamW": 1.9221635777617057}, "class": "stall", "probe_best": 1.916091866268607} +{"family": "jordan3", "seed": 601006, "x": [-0.058272399290189625, -0.24667086570729477, -0.5056371643022823, 0.10861159309549395, 0.5530049647880688, 0.09669716406117357, 1.8910284432628248, -0.4196952819142642, 0.5320155302718628, 0.20597203563423938, 1.7786414912184654, -0.3041298788375572, -0.19762810140114728, 0.06875203943871755, -0.01576694914713178, -0.04540178365528019, 1.193086481751724, -0.34010404988762577, -0.015958902613418535, -0.14112411013900394], "ratio96": 1.9967622641143539, "ratio_final": 1.9967694422831415, "diag": {"ratio": 1.9967694422831415, "nonnormal": 0.7262079146851976, "eig_gaps": [0.1963953942809951, 0.24103008207754498, 0.36396442879098645], "discness": 0.02453738116134489, "peaks": 4, "deg": 3, "diamW": 2.9088259411067434}, "class": "near2-other", "probe_best": 1.9967694422831415} +{"family": "jordan2", "seed": 602006, "x": [-0.1506024507409685, 0.40914758132244566, -0.002420090669959518, 0.32131974876931735, -0.1687762084565399, -0.03938403582392612, -0.016401330916424938, -0.18726583007842618, -0.885431664693575, -0.6916434265268065, 0.3318271640510078, -1.251433461149873, 0.169534319470325, -0.10243654801744398, 1.4618716408187167, 0.6148421760491166, -0.013422911406622887, 0.08139088718810736, 0.002595054926757693, 0.028550100840205227], "ratio96": 1.9930868777421036, "ratio_final": 1.993105993398295, "diag": {"ratio": 1.993105993398295, "nonnormal": 1.3056979898087209, "eig_gaps": [0.09724428195753732, 0.2242439393712989, 0.2534205531844505], "discness": 0.012580972357830272, "peaks": 3, "deg": 3, "diamW": 1.7713624478525372}, "class": "near2-double", "probe_best": 1.993105993398295} +{"family": "nearnormal", "seed": 603006, "x": [0.6806268242780293, 0.09231473659106554, 0.3057122064111115, 0.1204532306247112, 0.5499176738545962, -2.883504777139101, -0.6301183386572278, -0.2312969224665807, 0.10504234360177234, -0.06786019736996708, 0.07606461459652447, -0.029647648054681206, 1.6359228526233929, 1.743349170045624, -2.87755640610899, -3.9379571676150906, -1.6909557689230361, 3.276671009011423, 0.7516185441950949, 0.18335164369143725], "ratio96": 1.7037919884708002, "ratio_final": 1.605585731404586, "diag": {"ratio": 1.605585731404586, "nonnormal": 0.09950905800750735, "eig_gaps": [0.11293950794617383, 0.8947854058556918, 0.9053530882057428], "discness": 440027496798.232, "peaks": 2, "deg": 3, "diamW": 3.328942033456778}, "class": "stall", "probe_best": 1.7044467150628857} +{"family": "companion", "seed": 604006, "x": [0.5225441605490881, -1.7613118650685273, 0.5889496596213541, -1.4145887059457458, -0.4555675517706028, 0.9163087468313333, 0.8630803906648437, 0.3616335483373272, 0.0287224204908321, 0.011138946787502903, -0.02507645986131749, -0.00996487455186671, 1.696532736056071, -0.7922295732070382, 0.93991867984463, 0.6074034330103233, 0.1752653487754826, -0.4432628126777405, 0.3915871740211978, -0.00213731260614041], "ratio96": 1.7608217278803144, "ratio_final": 1.6609233832027015, "diag": {"ratio": 1.6609233832027015, "nonnormal": 0.17436249184411418, "eig_gaps": [0.11059798413867107, 0.800206445229223, 0.8930785433713792], "discness": 0.8532135470866169, "peaks": 1, "deg": 3, "diamW": 3.191965810420953}, "class": "stall", "probe_best": 1.798108219610819} +{"family": "peaks", "seed": 605006, "x": [-1.1130529289735753, -0.42256868589262486, 0.06865363037388733, -0.28214166672309005, -0.42574303272815783, 1.2680048604354237, 1.130126793657418, -2.1441003054207544, -0.14653098866284386, -0.3174489608639708, 2.7190705863778666, 0.3109041739793869, 2.192517909994469, -1.5243217406338299, 3.4135845831105303, -1.6543503295729602, 3.0137166931889006, 0.37097690285095586, -0.3396156462311285, -0.19936288286293907], "ratio96": 1.9529974794233007, "ratio_final": 1.9532009263603356, "diag": {"ratio": 1.9532009263603356, "nonnormal": 0.7388439986634598, "eig_gaps": [0.2884495705104577, 0.3943878967982939, 0.44234961146453], "discness": 0.03801425528039115, "peaks": 3, "deg": 3, "diamW": 4.125577531573073}, "class": "stall", "probe_best": 1.9679453455622358} +{"family": "real", "seed": 606006, "x": [2.194078930974954, -0.16133264171300518, 1.8663653863386307, 0.04018776700677672, -1.3028342598008893, 0.07698918129032617, 0.8803405597072707, -0.09595124729347422, -0.10733887224304021, 0.20660247493305264, 0.22205517293805097, -0.04835076332298455, 1.8178047118634726, 0.10085545124698087, 1.5708551114283686, 0.19023216332283968, 1.0030627804084804, -0.01556701391891098, -1.0045349880988765, -0.1259896812145136], "ratio96": 1.6876085032318193, "ratio_final": 1.6270231550005925, "diag": {"ratio": 1.6270231550005925, "nonnormal": 0.17229299544190277, "eig_gaps": [0.10071952536305223, 0.8297589486133367, 0.9176227405690216], "discness": 0.7429776497061936, "peaks": 1, "deg": 3, "diamW": 3.819679579380595}, "class": "stall", "probe_best": 1.6788835981568102} +{"family": "random", "seed": 600007, "x": [0.19050876180930423, -0.33931874149884944, -1.450342740785822, 0.11763006911247292, -0.147760397844507, 0.37224454316947386, 0.09076127105878186, -1.7553425230938853, 0.2733228254745358, -1.6220548717636376, 1.7762550993899886, -0.24035271422192084, -0.0172844485381676, 0.46773749840058526, 1.4560867387536074, 4.003179114732218, 0.9959617126324976, 2.987779312369618, -0.1694745131050691, -0.8516145209982313], "ratio96": 1.884040897673235, "ratio_final": 1.8869796101928276, "diag": {"ratio": 1.8869796101928276, "nonnormal": 0.7581064592742784, "eig_gaps": [0.22992967111891272, 0.3873331227844364, 0.4970793966779462], "discness": 0.04892745309412353, "peaks": 3, "deg": 3, "diamW": 3.4265948881404276}, "class": "stall", "probe_best": 1.964297443278889} +{"family": "jordan3", "seed": 601007, "x": [0.6602763872159385, -0.04212155390755326, -0.38148330407046827, -0.2495819158142698, -0.29484198858187277, -0.26671482927774925, 1.6008855898268655, 0.3317634854552758, -0.20671584687138825, 0.6338557025317748, 1.8741680282116966, 0.28830410037831444, -0.4187311646252495, 0.017499290805793745, 0.11757974688769009, 0.553090682330313, 1.2532147189119085, 0.08931471803867097, 0.04464402086294145, -0.19521368363255243], "ratio96": 1.7493161961282113, "ratio_final": 1.8974921440997967, "diag": {"ratio": 1.8974921440997967, "nonnormal": 0.8476050672494106, "eig_gaps": [0.03119361581858506, 0.3465415412889638, 0.37516664649464143], "discness": 0.05769357681048524, "peaks": 3, "deg": 3, "diamW": 2.831318071765534}, "class": "stall", "probe_best": 1.900197903943355} +{"family": "jordan2", "seed": 602007, "x": [-0.1755971928768279, 0.030048183450366556, -0.14410604817174472, -0.0035746936720660186, -0.19954201725929563, -0.08346223405702496, 1.2678346521186104, 0.3174409369367081, 0.027628912172930672, 0.1524371130868865, -0.004474898098319281, 0.015771080200343637, 0.11848530204303681, -0.1906390403508627, 0.6907080965425434, -1.126135678052263, 0.02336367605060407, 0.04943586533287095, -0.003238865062849223, -0.0008477517386126752], "ratio96": 1.9987495711606287, "ratio_final": 1.998750459681895, "diag": {"ratio": 1.998750459681895, "nonnormal": 1.3382640387073814, "eig_gaps": [0.034977466959736025, 0.07382964680278367, 0.08808176057979718], "discness": 0.0003257586144565633, "peaks": 3, "deg": 3, "diamW": 1.317054610838134}, "class": "near2-triple", "probe_best": 1.998750459681895} +{"family": "nearnormal", "seed": 603007, "x": [-1.0513391630496558, -1.6959262119045477, 1.7186235004831798, 1.7915237366076406, -0.8367520905874531, -0.38497179063937015, -1.1961018030259693e-05, 8.96186592711346e-06, -0.021760220352189514, -0.1533253353863978, 1.0933630902174662e-05, -1.5484994447905858e-05, -0.013340490229108075, 0.12788299057773408, -0.4526766173311968, -0.1882013624076047, 0.2613861914994503, 1.1664718536555214, 1.110828842770521, 0.023993327994260932], "ratio96": 0.9999999999727042, "ratio_final": 0.9999999999797999, "diag": {"ratio": 0.9999999999797999, "nonnormal": 0.026586097386214255, "eig_gaps": [0.29792812482622527, 0.7528144692765273, 0.9988457557940946], "discness": 0.5328967851020309, "peaks": 1, "deg": 3, "diamW": 4.458796762420723}, "class": "one-exact", "probe_best": 0.9999999999806564} +{"family": "companion", "seed": 604007, "x": [-0.7437880378231271, 1.0143845499147166, -0.8243501636679489, -0.8556911519520369, -1.1760441395784644, -0.6864533086341797, 1.2168401818021288, -0.1296392136775943, 0.6707746122276355, -0.3255476336974368, 1.1439339432255933, 0.03397712589622506, -0.8074943892469328, -0.25307581691281766, -0.12511933452404866, -0.3314547812205869, 0.5631564589687239, 0.12430382115701638, 0.6930445827479801, 0.09554675134526147], "ratio96": 1.6497401469219524, "ratio_final": 1.6494641836904647, "diag": {"ratio": 1.6494641836904647, "nonnormal": 0.5853250747270698, "eig_gaps": [0.1421379117609941, 0.6391030697693364, 0.6816773654077716], "discness": 0.4487763205000882, "peaks": 4, "deg": 3, "diamW": 2.745888722199415}, "class": "stall", "probe_best": 1.6536512069429743} +{"family": "peaks", "seed": 605007, "x": [-0.1765898024295985, 0.3973552834877251, 0.031397974038863394, -0.3680816797706945, -0.9447651964782122, -0.6517984832669272, 4.211456028591624, -4.300165444682208, -0.6071026710009543, -0.3315498298007954, 5.937619270454003, -0.4255896950705138, -0.10074921238966414, 0.48470500207053185, 2.5086271156891238, 1.85145439104362, 3.3796711949965075, 0.08236626968640567, -0.026033216815534145, 0.07051343562158108], "ratio96": 1.9983515309414068, "ratio_final": 1.9983602386976598, "diag": {"ratio": 1.9983602386976598, "nonnormal": 0.7115370492397998, "eig_gaps": [0.0924183005062884, 0.1184437192826544, 0.1515054550959211], "discness": 0.002737086470106558, "peaks": 3, "deg": 3, "diamW": 8.582623087335197}, "class": "near2-double", "probe_best": 1.9983602386976598} +{"family": "real", "seed": 606007, "x": [-0.6151886362426193, -0.08152155331772727, -0.6606707545365024, 0.1317814011820978, -0.06123834051670721, 0.15066678146271056, -0.06774711704762004, -0.07914521744676505, -0.5354278445369385, 0.25945135223097765, 1.8382864684430495, 0.25997265467715924, -0.6701710879328178, 0.29518397824760245, -1.9243637927153379, 0.02214011678336568, 0.030793185360431838, -0.017331491726974152, 0.04390353328513642, -0.001342677952164142], "ratio96": 1.9958947807005225, "ratio_final": 1.9963184216629908, "diag": {"ratio": 1.9963184216629908, "nonnormal": 1.2023536259774834, "eig_gaps": [0.10686893498297914, 0.29387002500016596, 0.29431761734762635], "discness": 0.02623759878684691, "peaks": 4, "deg": 3, "diamW": 2.0407996232833407}, "class": "near2-double", "probe_best": 1.9963184216629908} +{"family": "random", "seed": 600008, "x": [0.5482634954677115, -0.299567343125135, 0.8339297211755294, 0.6846719204862453, 2.002667724099597, 0.4695714410048557, -1.2153104003004334, 0.6787635599404432, -1.7382846966071155, 0.23154728149793855, -0.734200679651765, 0.03874989226563717, 1.0575939649486854, 0.5208838254262913, -1.1000389030809221, -0.22719606738642104, -0.10673889806581191, 0.17549066376895378, 0.03973983992577486, -0.042670478937061135], "ratio96": 1.8890659814715745, "ratio_final": 1.8890855571076424, "diag": {"ratio": 1.8890855571076424, "nonnormal": 0.822547496930886, "eig_gaps": [0.3595429669904677, 0.41690600573691644, 0.5771927447176084], "discness": 0.21483406065415805, "peaks": 3, "deg": 3, "diamW": 2.850444080698576}, "class": "stall", "probe_best": 1.9534811158564258} +{"family": "jordan3", "seed": 601008, "x": [0.3144187325665587, -0.18575203670939486, -0.21504335122552581, -0.10853784669968543, -0.43052648678479544, 0.08600549270520569, 1.2732475711446907, 0.17279164951200213, 0.24460456204068093, -0.19125320291892253, 1.0833282424002353, -0.5822306288580158, -0.06766390234332091, 0.09639171599360932, 0.3572780724956082, 0.22675233234629627, 1.5053521403353063, 0.15566351206502066, -0.226102968952146, -0.006682338925734668], "ratio96": 1.988205468037788, "ratio_final": 1.9881820015402822, "diag": {"ratio": 1.9881820015402822, "nonnormal": 0.7227059729623553, "eig_gaps": [0.14732621988944067, 0.2715327687653453, 0.40241324644991006], "discness": 0.03340913786013266, "peaks": 3, "deg": 3, "diamW": 1.9705273011058309}, "class": "near2-double", "probe_best": 1.9881820015402822} +{"family": "jordan2", "seed": 602008, "x": [-0.28405817948517365, 0.1806493940053815, -0.3944030929736102, 0.1866288220882787, -0.36314847976788145, 0.20587910802983544, 0.3543830947735134, 0.13520670188055783, -0.041968852759149994, 0.0273578958061311, -0.015532622208690394, 0.009151275599596543, 0.5327853667845723, -0.1214754465022394, 0.8512062560561802, 0.3144321463322617, -0.5675158180895041, -0.31720865100516726, 0.3734599255196807, 0.6905925565015956], "ratio96": 1.9372081271870614, "ratio_final": 1.9372096446727198, "diag": {"ratio": 1.9372096446727198, "nonnormal": 0.34381499487903494, "eig_gaps": [0.09215199820703723, 0.20841037563267303, 0.27742243207001877], "discness": 2680711282.6397796, "peaks": 2, "deg": 3, "diamW": 0.3983340572052407}, "class": "stall", "probe_best": 1.9397633804367667} +{"family": "nearnormal", "seed": 603008, "x": [0.4406406908494128, 0.9952198993497483, 1.30036157648349, -0.6470946330921177, -0.7481852834351249, -1.6493604991910074, -0.834272157715913, 1.3998340575693882, 1.6183664322634361, -5.694141860204088, -1.1810542330392777, -0.5308794939238032, 0.12510015552246123, -0.5076516774084071, -1.2755877873408201, 7.043141987809309, 0.1486449499869461, 0.025131633622950747, -0.02158999908884418, 0.003116592568198883], "ratio96": 1.9667115082177578, "ratio_final": 1.9670049320670957, "diag": {"ratio": 1.9670049320670957, "nonnormal": 1.340063315758835, "eig_gaps": [0.26765939925503784, 0.3292931526298168, 0.41865792939732044], "discness": 0.03081778349759919, "peaks": 4, "deg": 3, "diamW": 6.925706979684686}, "class": "stall", "probe_best": 1.9861643329106857} +{"family": "companion", "seed": 604008, "x": [-0.2754563175082258, 0.17265408307331592, 1.3536145022884947, 0.8097190008367943, 1.309742074114017, 0.8233509571289479, -0.0012833211786059755, 0.024160880543536163, -0.020601971227375247, 0.007771326074972624, 1.2021600453532069, 0.009956727647255451, 0.11351582909117386, -0.6837486465075397, -0.2161108200227273, -0.26914960406275357, 0.46074031992624387, 0.10020890047051029, -7.127256550480243e-05, 0.09217994967042087], "ratio96": 1.7672376458983716, "ratio_final": 1.622248131205978, "diag": {"ratio": 1.622248131205978, "nonnormal": 0.3182435267271336, "eig_gaps": [0.019692726214498957, 0.7345106217806954, 0.7497938716187902], "discness": 229741440081.4133, "peaks": 1, "deg": 3, "diamW": 2.3329165321236696}, "class": "stall", "probe_best": 1.7972755099436621} +{"family": "peaks", "seed": 605008, "x": [-0.7477112935789108, -0.28358036524013674, -0.4678700580987505, -0.21884097137186664, 0.9261623693922111, 0.17081245531790867, 1.1892311176438057, 0.0001253994736189667, -0.018636355851219838, -0.011159107011227073, 0.018325474671064525, -0.016141605181029212, -0.6574223280087002, -1.8152894798197725, -1.2602298372620813, -0.9922060938248987, 1.4962512061315874, 1.4662287094996533, 0.5882373003137391, 0.12182396713467504], "ratio96": 1.8380817392523627, "ratio_final": 1.744658482946865, "diag": {"ratio": 1.744658482946865, "nonnormal": 0.641763085196776, "eig_gaps": [0.1303888057630317, 0.6570757912943254, 0.7873533921528832], "discness": 0.3484063465543013, "peaks": 1, "deg": 3, "diamW": 2.202889620666029}, "class": "stall", "probe_best": 1.8596839192268524} +{"family": "real", "seed": 606008, "x": [-2.622064035328334, 0.0571702830214681, 0.542352880121455, -0.01413354701221095, 0.7690079284825266, -0.03242538330513289, 8.435894367132835e-06, -1.4193304733438171e-05, -1.2779171572293208e-05, -8.64223751138383e-06, 0.4881015858949144, -0.11579077576896114, 0.46050241676982806, 0.20003761566211614, 0.9156377744261623, -0.1189037599114616, 0.6310289659164243, -0.0799183730878677, -0.8139797690448387, -0.13142543325965544], "ratio96": 0.999999999965622, "ratio_final": 0.9999999999771068, "diag": {"ratio": 0.9999999999771068, "nonnormal": 0.048739896951026235, "eig_gaps": [0.06398110411835486, 0.8905956171817622, 0.9544763420564613], "discness": 1.3841769024292512, "peaks": 1, "deg": 3, "diamW": 3.554048658343728}, "class": "one-exact", "probe_best": 0.9999999999775342} +{"family": "random", "seed": 600009, "x": [-0.2164169344209485, -0.019747824112128887, -0.3008146097829501, -0.41085406844582595, -0.3571856208042723, -0.01670345958414328, 0.5305050817572354, -0.22080049178383426, 0.040928195294475694, 0.8571630482677889, 0.5509943936530065, 0.4223772041132204, 1.5565260931510856, 0.1821270504997255, 2.8850108727434534, -2.6685427496953396, 1.1192857391848252, 0.4762969663055831, 0.302847829450731, 0.09892541670860383], "ratio96": 1.9442929439253747, "ratio_final": 1.9457992572343816, "diag": {"ratio": 1.9457992572343816, "nonnormal": 1.1435837574526493, "eig_gaps": [0.1082457621860932, 0.30609928396757774, 0.30759652350093303], "discness": 0.03445977897098319, "peaks": 3, "deg": 3, "diamW": 1.300758565474143}, "class": "stall", "probe_best": 1.9691451896446446} +{"family": "jordan3", "seed": 601009, "x": [-0.15452597856081968, -0.2839683438384515, 0.09300237593242727, 0.09847117048060361, -0.03996682658513602, -0.07811844713263572, 0.888890671325858, 0.47327520700076803, -0.1558962019492356, -0.0748215395879191, 0.8034979112159288, 0.6845926427845052, 0.00047005054978060197, -0.006927384464892248, 0.07800210973077021, 0.2622009402706389, 1.0474364152827778, 0.20555117948146984, -0.11728884255402644, 0.12751776275518656], "ratio96": 1.9930046020093364, "ratio_final": 1.9930006270504441, "diag": {"ratio": 1.9930006270504441, "nonnormal": 0.7024397595631039, "eig_gaps": [0.14378460892338557, 0.15323337214627833, 0.2963162696688254], "discness": 0.018006644699606302, "peaks": 3, "deg": 3, "diamW": 1.5373941570266718}, "class": "near2-double", "probe_best": 1.9930006270504441} +{"family": "jordan2", "seed": 602009, "x": [-0.19452431237261925, 0.09804837037695308, -0.008373245083055543, 0.2984988326192781, -0.309551064039311, 0.2748523537743246, 0.20590487262715007, 0.2790996164689349, 0.5365513785429842, -0.18216361816799043, 0.6701366491913361, -0.1831037259945752, 0.08436287406992354, -0.2215751657254243, 0.6373239593855158, -0.002993022550323708, -0.06283572381758612, 0.0026877576193868307, -0.11423064119124599, -0.030049259237089287], "ratio96": 1.9691040387895242, "ratio_final": 1.9692063347728552, "diag": {"ratio": 1.9692063347728552, "nonnormal": 1.0711010143670976, "eig_gaps": [0.2099987158342068, 0.27234973725392303, 0.30077308254592383], "discness": 0.07522100197009585, "peaks": 4, "deg": 3, "diamW": 1.0044272324409476}, "class": "stall", "probe_best": 1.9700121237175372} +{"family": "nearnormal", "seed": 603009, "x": [-0.14125951240596646, -1.045088897498958, 0.21486123972053234, -1.1963393869777597, -0.24984627733173906, 2.0516673950431725, -0.4115836268199198, -0.2220817349609956, 9.182810360824608e-06, 6.78951835647619e-06, 5.45644495742951e-06, -1.4128577640834595e-05, -0.052495175927409585, 0.0066326006355122815, -0.1385833654375104, 0.268051571467903, 0.38430424135099617, 1.0186433194489815, 0.9643854846529082, 0.13099834458989384], "ratio96": 0.9999999999705363, "ratio_final": 0.9999999999813017, "diag": {"ratio": 0.9999999999813017, "nonnormal": 0.05670174667421663, "eig_gaps": [0.11257282322987744, 0.9015678092679277, 0.9546444851364765], "discness": 0.5686714704213492, "peaks": 1, "deg": 3, "diamW": 3.4369677531509044}, "class": "one-exact", "probe_best": 0.9999999999825987} +{"family": "companion", "seed": 604009, "x": [1.2922586602296087, 2.7707623149437524, 2.9125009240157507, -0.37938020237817804, -3.113706695687682, 1.545549517831895, 2.475753327274261, -0.9667452291793286, -2.1884973533737884, 0.7152035096887757, 3.156829427079924, 0.10014328876174808, -10.94925612077938, 1.8758524019763825, 2.0856689459588265, -6.057786270380361, 3.2893842322686746, 1.48974287416753, 0.1226484233609037, -0.0037053112865519755], "ratio96": 1.638834562593091, "ratio_final": 1.6488330511479956, "diag": {"ratio": 1.6488330511479956, "nonnormal": 0.6534134011379263, "eig_gaps": [0.4494054594797422, 0.5801712472833456, 0.8025691003900688], "discness": 0.1906597734634174, "peaks": 3, "deg": 3, "diamW": 7.882409579119808}, "class": "stall", "probe_best": 1.6499348909728633} +{"family": "peaks", "seed": 605009, "x": [-0.9770454118459148, 0.5789408300213066, -0.1661107948295532, 0.08098090923521865, -0.009381243077366484, 0.5770246055507922, 1.2312652198812881, -0.08006190456450783, -0.31406852558517645, -0.03930833011110445, 0.11024209482619048, 0.0886714697952103, -0.20501751380853106, 1.9422176938266977, -1.3004541738138444, 1.8852716626047639, 1.3317971907853434, -0.12113918848090645, 0.5958310778248777, 0.1889152992556235], "ratio96": 1.9213514804747929, "ratio_final": 1.923676024962603, "diag": {"ratio": 1.923676024962603, "nonnormal": 0.8649163526050839, "eig_gaps": [0.3269720448734276, 0.5981237524528749, 0.6082097541344051], "discness": 0.9564202800672704, "peaks": 5, "deg": 3, "diamW": 1.5910071476131389}, "class": "stall", "probe_best": 1.924194902871596} +{"family": "real", "seed": 606009, "x": [-0.5689875963800747, -0.18222037832968788, 1.0323002692075458, -0.456675383666695, 1.4205340104631095, -0.42657549581237353, 0.00941851681748685, -8.756801144055837e-05, -0.0039375193445496965, 0.008208539000465599, -1.338891670585358, -4.451387435844728e-05, 3.324988887650994, -0.841958699280823, -0.22777631602393977, -0.7560910115971269, -2.2654118649659747, -0.18837802480073096, 0.40368039163885605, 0.10408042294737152], "ratio96": 1.8785476750536976, "ratio_final": 1.6579877364104212, "diag": {"ratio": 1.6579877364104212, "nonnormal": 0.46951892266010653, "eig_gaps": [0.15513918229279236, 0.6472669812239553, 0.7985952703625642], "discness": 1.9861807191220506, "peaks": 1, "deg": 3, "diamW": 2.5099966001417213}, "class": "stall", "probe_best": 1.924673574388366} +{"family": "random", "seed": 600010, "x": [-0.9647657048168763, 0.1392075826832031, 0.7627283799601545, 0.6685157873257976, 0.5466804302019641, -1.2198553410155584, -2.600587905310131, 1.4474814654757742, 1.276208665006937, -0.941381438603377, 2.2694837775804797, 2.1260996088889943, -0.9561468614661242, -0.8130358606492252, -1.4981896804376529, -1.3456689135179758, -1.0719616363257494, 9.286292965470945, 0.5823426512417014, 0.2239526101372098], "ratio96": 1.9429627572578805, "ratio_final": 1.943184189433361, "diag": {"ratio": 1.943184189433361, "nonnormal": 0.7411167999888572, "eig_gaps": [0.3600297208724793, 0.3787457303467211, 0.40503399563422265], "discness": 0.012665356300774084, "peaks": 3, "deg": 3, "diamW": 5.018379837105981}, "class": "stall", "probe_best": 1.9537770906860346} +{"family": "jordan3", "seed": 601010, "x": [0.5904444312723007, -0.08024797934183736, -0.12132562466026747, -0.16176682695999922, -0.4647740918811282, -0.15554204454529585, 1.1135816378252572, -0.08115981571009254, 0.04586179742212153, -0.09520779972082072, 1.087353356769134, -0.5716570074524002, -0.26791728337751375, -0.030865864613020877, -0.08042073664650118, 0.4003510776859256, 1.7391294756385638, -0.03913508183805342, 0.10846030908972132, -0.060498110273579486], "ratio96": 1.977725341824112, "ratio_final": 1.977700890059782, "diag": {"ratio": 1.977700890059782, "nonnormal": 0.6877018385727185, "eig_gaps": [0.17322457083249462, 0.3612818303676745, 0.533484462524563], "discness": 0.07194895273053578, "peaks": 4, "deg": 3, "diamW": 1.9830031676546678}, "class": "near2-other", "probe_best": 1.977700890059782} +{"family": "jordan2", "seed": 602010, "x": [0.27929163963725523, -1.081113637630387, 0.17306812483781364, -0.48256335045599874, 0.43218601227519327, -0.47993174971118924, 1.1675685630610204, -0.09297288864977585, 0.6468213314331627, 0.2563949110376399, -0.12964184097222298, -0.14417053595525875, -0.23553825728809508, 1.5806588224409692, 1.6082872591610675, -0.2266093231821642, 0.13102842182725138, 0.48448155763407885, 0.1524924640021304, -0.09897354790871543], "ratio96": 1.9828324112171216, "ratio_final": 1.9829410234617277, "diag": {"ratio": 1.9829410234617277, "nonnormal": 0.7553613045645912, "eig_gaps": [0.1729015570179441, 0.40561433650324774, 0.41389919946457515], "discness": 22710183687.991165, "peaks": 4, "deg": 3, "diamW": 1.4987213234399015}, "class": "near2-other", "probe_best": 1.9829410234617277} +{"family": "nearnormal", "seed": 603010, "x": [-0.6093176821268349, -0.28684300635235593, 1.1555752897359537, 0.93970200953181, 0.6863889043002778, 1.8258550889360414, 0.9351933438147279, 0.33388682250426444, 0.10157652740274954, 0.5759301060623156, -1.6256294283431778, -1.0586700613167288, -1.350220452584944, 0.7937964281321792, -4.9470227412257195, -0.8004919098319574, 1.9901669261720123, -3.4585486486898542, 0.37253431117680447, 0.3889819682781135], "ratio96": 1.8265228027520146, "ratio_final": 1.8307680805342925, "diag": {"ratio": 1.8307680805342925, "nonnormal": 0.5616452305243439, "eig_gaps": [0.2890270807613838, 0.6195189781591827, 0.7143904729507815], "discness": 0.4753444182842701, "peaks": 3, "deg": 3, "diamW": 3.4692179367615994}, "class": "stall", "probe_best": 1.9188516491727974} +{"family": "companion", "seed": 604010, "x": [0.32541938807110826, -0.7896395486610123, 1.128309187652112, 0.3712361326262055, 2.6916881978551856, -0.5049974516483481, 1.511355672227492, 0.3925218101766922, -0.09999899257959445, 0.07999440556052599, 0.2100223434622077, 0.20996386340981926, -0.2975914393258991, 0.3182634687769481, -0.13135990844478299, 0.09033756296046282, 0.7068648534278181, -0.05844877228577516, 0.9746660911347274, -0.03782453818242493], "ratio96": 0.9999999999740526, "ratio_final": 1.0027249091450632, "diag": {"ratio": 1.0027249091450632, "nonnormal": 0.3815171986764045, "eig_gaps": [0.49267072732404044, 0.6255570652200253, 0.8318920400010905], "discness": 194814427293.44205, "peaks": 1, "deg": 3, "diamW": 2.864947882815614}, "class": "stall", "probe_best": 1.0042830480829974} +{"family": "peaks", "seed": 605010, "x": [-0.9320539125737016, 0.14053406042985958, -0.23030552286095932, -0.02481341023777415, 0.7058953164050452, -0.8931076858592911, 1.9336263005424366, -0.030903682458190362, 0.0033064041077288447, 0.44123017016606536, 1.6941423955083896, -0.285932050939178, -1.3943829669014896, 1.685723066873059, 0.22819379759063463, 2.6934558303377414, 3.568422137914287, 1.100451344487354, 0.1768570568214559, 0.10487056835788253], "ratio96": 1.9538452530896613, "ratio_final": 1.954813825118366, "diag": {"ratio": 1.954813825118366, "nonnormal": 0.6910917299827187, "eig_gaps": [0.21989912896923347, 0.38945502664403614, 0.590744644108898], "discness": 0.10321811476562459, "peaks": 4, "deg": 3, "diamW": 3.278617078270614}, "class": "stall", "probe_best": 1.9585307633534743} +{"family": "real", "seed": 606010, "x": [-0.6686592052778355, 0.03385692084974145, 1.1094219045118678, 0.33570108235098317, 0.7053445338321562, -0.2438736725531393, -3.1503493680630386, -0.5231458638000837, -2.3703533659846805, 0.09060084619947861, 0.06316201088904805, 0.2955441535848159, 0.22392191301683706, 0.22784969215175627, -2.6950458530766848, 0.32505489340198657, 0.034664744957582594, -0.04976159538227004, 0.022215673384456597, -0.01165184371373916], "ratio96": 1.980061967603888, "ratio_final": 1.9801538184736822, "diag": {"ratio": 1.9801538184736822, "nonnormal": 1.3386684041387473, "eig_gaps": [0.16242788027490174, 0.3222650293962243, 0.4146201999450412], "discness": 0.03330029299797145, "peaks": 3, "deg": 3, "diamW": 4.349810813006577}, "class": "near2-other", "probe_best": 1.9801538184736822} +{"family": "random", "seed": 600011, "x": [0.8009286265553185, -0.35446897951504064, 0.07440495143561181, -1.1365597863955927, -0.7836667242539859, -0.25521229230899856, 0.3304005056466759, 1.886074756051021, 0.08528501463876817, 0.9625671743430814, -0.8601619564969181, 1.6475845176709862, 0.013366787779031976, -1.59936676717848, -3.17962889185179, -0.13690403607100038, -0.14383960830048062, 2.431722886943943, -0.3280384002869576, 0.06166203197400973], "ratio96": 1.9124954869643978, "ratio_final": 1.91721205478952, "diag": {"ratio": 1.91721205478952, "nonnormal": 0.6897229123400266, "eig_gaps": [0.3287242908534352, 0.378793270523012, 0.4889256558302207], "discness": 0.04511909418720725, "peaks": 3, "deg": 3, "diamW": 3.2473259153877763}, "class": "stall", "probe_best": 1.9666137638510621} +{"family": "jordan3", "seed": 601011, "x": [-0.1403988355693968, 0.361980362883084, -0.10796959397809319, 0.28981575938266046, -0.0730112933009956, 0.4271948775030475, 0.24373332679315873, 0.2560565375883579, 0.023267637084623406, 0.10700316143982164, 0.011287871926688356, -0.03550339951293588, 0.18104507325772745, 0.4749574765019463, -0.6993261563487687, 0.6552296456085518, 0.9153196310220455, 0.4901782388929077, 0.4571817573865397, -0.019930979351225814], "ratio96": 1.969419348853536, "ratio_final": 1.9694424988248191, "diag": {"ratio": 1.9694424988248191, "nonnormal": 0.35039223360605026, "eig_gaps": [0.2075952312785431, 0.24606230427486614, 0.3719603559484212], "discness": 1435906517.4969554, "peaks": 3, "deg": 3, "diamW": 0.38110836737956805}, "class": "stall", "probe_best": 1.9760591249943464} +{"family": "random", "seed": 100000, "x": [-0.4693482384723695, 0.5606797722517263, 1.22528817159415, -0.43567083789954375, -0.6795273738738636, -0.30378262430531733, -0.6297257982192327, -3.5249875132345547, 0.18851235213902695, -0.9064018304154038, -3.448427192307559, 0.28016577576209023, -1.3899388091492442, 0.826307405356063, 1.0374241682979481, 0.08188603174201664, 4.783392829889163, 0.9044251590751338, -0.25973072926373575, -0.11733388971453573], "ratio96": 1.9730777171334675, "ratio_final": 1.9730790509888327, "diag": {"ratio": 1.9730790509888327, "nonnormal": 0.7137613572322133, "eig_gaps": [0.16159029350180465, 0.3468082060913677, 0.3570630886043997], "discness": 0.025587354690489943, "peaks": 3, "deg": 3, "diamW": 5.505567574650626}, "class": "near2-other", "probe_best": 1.9730790509888327} +{"family": "random", "seed": 100001, "x": [0.5203954203589413, 0.4186912687683316, 0.782607554009086, -0.6918064853130799, 0.3618122405963487, -0.6530564709920417, 1.1680238904544085, 1.5904475709165093, 0.9070337387321544, 0.18629165326981678, -0.2011676894440021, 1.8896963315180075, -0.22291182448109179, -1.8574186717159689, -0.7777280120589901, 3.678208044467908, 1.6764566763704514, -1.9131281040921406, 0.4273480301605477, 0.004481990210881113], "ratio96": 1.9453020924988285, "ratio_final": 1.9454260507662304, "diag": {"ratio": 1.9454260507662304, "nonnormal": 0.684407907980499, "eig_gaps": [0.13459303878812942, 0.3450750640888819, 0.3634267849831733], "discness": 0.04264622441053452, "peaks": 3, "deg": 3, "diamW": 3.139655273293216}, "class": "stall", "probe_best": 1.9578107851896638} +{"family": "random", "seed": 100002, "x": [-0.10761593427850549, -1.1825688711613145, -0.12947956518585124, -1.3593847517763082, -0.04927260230867228, -1.0821474541347444, 0.6149112783994763, 0.07992189594406728, -0.06863344608840112, 0.21180176695142008, 0.12964171232901778, 0.015634019224478768, 0.7182466927553433, 2.35759862560332, 0.5920982415051725, -0.6802454290243981, -0.4323296167326621, 0.7183647369494864, -0.1463951815967059, 0.0774973592275475], "ratio96": 1.9658337699795976, "ratio_final": 1.9658338189373585, "diag": {"ratio": 1.9658338189373585, "nonnormal": 0.1337537103673971, "eig_gaps": [0.16789234009972617, 0.2575531459870722, 0.41721180889533344], "discness": 3437866816.164902, "peaks": 3, "deg": 3, "diamW": 0.6917504057179872}, "class": "stall", "probe_best": 1.9664291681370298} +{"family": "random", "seed": 100003, "x": [-8.539297932575458, 4.662726037460518, 0.07273790925563822, 1.028118900236058, 0.8090269418860376, -1.0258483067960487, -4.706364470695651e-06, -1.99399966615987e-06, 1.938701010736351e-06, -1.3557192736694334e-06, 1.6560700360789555, -0.08602661168368617, -0.7991828140297721, 0.548197300771383, -0.9107805926093497, -0.33098929554129014, 1.058014039495844, 0.07710195359262512, -1.1590147625157772, -0.37939946345629766], "ratio96": 0.9999999999653804, "ratio_final": 0.9999999999999278, "diag": {"ratio": 0.9999999999999278, "nonnormal": 0.0641569723312452, "eig_gaps": [0.1933872324441422, 0.8284818956729327, 0.9698907458808456], "discness": 1544420728328.4534, "peaks": 10, "deg": 3, "diamW": 11.28279923451957}, "class": "one-exact", "probe_best": 0.9999999999999294} +{"family": "random", "seed": 100004, "x": [0.3961471874611877, -1.649764809556161, 0.1672914997542405, -1.0533666284966348, -0.11917853536206535, -0.6382581813903143, -1.0685708267423633, -0.4861571894849772, 0.1375283021699676, 1.0367410569747728, -0.1307395501674027, -0.08212472736410398, -3.7101178302324413, 3.0384205362847645, -0.11469673241089628, 1.388820354855394, 0.20378629808110682, 1.9260164713962622, 0.08590739485707609, -0.23541741453467846], "ratio96": 1.7574423362270066, "ratio_final": 1.8343911617633484, "diag": {"ratio": 1.8343911617633484, "nonnormal": 0.5804259291457785, "eig_gaps": [0.26797355845505916, 0.33940280529035566, 0.603152791494703], "discness": 54277044991.53609, "peaks": 3, "deg": 3, "diamW": 1.8821300419546991}, "class": "stall", "probe_best": 1.8399996144383217} +{"family": "random", "seed": 100005, "x": [-1.4935464611800648, -1.73059755527555, 1.6988864698098056, 0.8565570954112128, 1.4315053066127517, 0.7390262001374228, 0.1499891499775842, 0.15001988462705437, 0.16001175879908608, 0.1600181422866594, 1.0093312375145804, 0.6182686261929704, 0.5345175940123149, -0.7286601896235778, -1.0616364743443532, -0.06557360364574379, -0.11204433994459281, 0.2961641202380269, 2.1046944054042216, -0.8165585901484685], "ratio96": 0.9999999999653757, "ratio_final": 1.0613982600645762, "diag": {"ratio": 1.0613982600645762, "nonnormal": 0.22342849408437326, "eig_gaps": [0.06367191502664943, 0.8345456092290574, 0.8957939315324392], "discness": 0.4957661357273988, "peaks": 2, "deg": 3, "diamW": 4.587144071392523}, "class": "stall", "probe_best": 1.0655608892730972} +{"family": "random", "seed": 100006, "x": [0.866855044325146, -0.8782257982727499, 0.8694654488006663, 0.7438511532544664, 0.7728125455024923, -0.5580193472333204, -0.0835897101312655, -0.02396449423394648, -0.34269307607572724, 0.655384484125632, -0.051425223061235655, 0.08221397858010786, 1.1672458472685376, -0.9064223272604035, -0.35907704609346, 0.0795080231557613, -0.8271017330839479, 0.5532455992277013, 0.9367687891956982, -0.4495051181781817], "ratio96": 1.7285252713844548, "ratio_final": 1.6810540159458285, "diag": {"ratio": 1.6810540159458285, "nonnormal": 0.2079840914399782, "eig_gaps": [0.17850449501027432, 0.698255564318281, 0.8676109943271771], "discness": 209493532389.32172, "peaks": 1, "deg": 3, "diamW": 1.8695925507981073}, "class": "stall", "probe_best": 1.741875846824156} +{"family": "random", "seed": 100007, "x": [1.408938420047168, 1.1049439721822487, 1.555098795046052, 1.2206975734502825, -0.4106576426173465, -1.0390088503111072, 0.30809928785882, -0.7919916947534265, -0.0769250078170034, 0.1504114454596065, 0.14825345234683962, -0.0351241121603172, 1.3473526681130714, 0.2329311015034383, 0.5258260119304556, -1.2883326464709006, 0.7811150853674307, -0.15559983368036057, 0.06863208464556098, 0.6948168958916812], "ratio96": 1.7695600072957667, "ratio_final": 1.5277950666639002, "diag": {"ratio": 1.5277950666639002, "nonnormal": 0.15122621461439986, "eig_gaps": [0.05575240106039937, 0.8408748239788189, 0.895612994251482], "discness": 1.076159926011765, "peaks": 1, "deg": 3, "diamW": 3.3441615498697104}, "class": "stall", "probe_best": 1.7601864253252406} +{"family": "random", "seed": 100008, "x": [-0.22535346106510534, -0.3246253418637256, -0.06354535724912982, 0.21343218316144996, -0.04033496507438562, -0.17374405756708577, -0.31972801764933667, -0.27348458130534875, -0.16866309483830194, -0.5160743851628061, -0.2848273856988516, 1.063846425164565, 0.1527454242650909, 0.2986503834268165, 0.2823036635391613, 2.669202164003307, -0.004301594057581123, -0.3690912972196131, 0.05369520437051117, 0.11686180572873747], "ratio96": 1.96941740382222, "ratio_final": 1.973967535275769, "diag": {"ratio": 1.973967535275769, "nonnormal": 1.256209215644697, "eig_gaps": [0.1790460479424658, 0.29088839184203125, 0.42137386147293165], "discness": 0.010760797276092134, "peaks": 4, "deg": 3, "diamW": 1.3334025506054474}, "class": "near2-other", "probe_best": 1.973967535275769} +{"family": "random", "seed": 100009, "x": [0.883549468394631, 0.7114839557841446, 0.3691278423747475, -1.048255324895143, -1.2058649234395178, 0.5054331497850711, 0.9658162013352815, -2.5089092841802305, 0.26163278564103043, -0.14782920263576088, -2.7833631324519548, -0.7666643893497079, 1.0055702398309105, -1.4460783754080409, 0.37140143049063756, 1.010145395127331, -2.130413012416887, 2.066669388453758, 0.3155536265255653, 0.2679355564425362], "ratio96": 1.9417827653610666, "ratio_final": 1.942288718514082, "diag": {"ratio": 1.942288718514082, "nonnormal": 0.7756922101391138, "eig_gaps": [0.40225826317346236, 0.4606560173733287, 0.48540802644036146], "discness": 0.02669466194273573, "peaks": 4, "deg": 3, "diamW": 4.5577388292963885}, "class": "stall", "probe_best": 1.9558347828983325} +{"family": "random", "seed": 100010, "x": [-1.5913379424991982, -0.369571518327837, -1.939944648634464, 1.5684761264231193, 0.4913866740332543, 1.7224500836102825, -1.625809391872662, -0.12771444289756198, -1.5313319631942086, -0.4053343557704905, 1.4245804918950022, -0.6921851399158407, 0.22872661691290896, 2.6828113442777153, -1.1997067174971006, 0.9781816671284244, 0.6782795893563587, 1.764677420391465, -0.3031157769865825, 1.2889929638141282], "ratio96": 1.479349774237704, "ratio_final": 1.4999138730193975, "diag": {"ratio": 1.4999138730193975, "nonnormal": 0.5493095038990424, "eig_gaps": [0.48457054312775727, 0.5995029124318217, 0.7264306642639918], "discness": 0.23201055709987173, "peaks": 3, "deg": 3, "diamW": 4.063703290614798}, "class": "stall", "probe_best": 1.5955972971505936} +{"family": "random", "seed": 100011, "x": [1.4626496829419486, 1.2946046115497407, -1.5554000634917475, 1.031956373329204, -1.3201002791530043, 0.8891186942290337, -1.5085136068958895, -5.746169054539017, 2.444995743724343, 0.4763911508213017, 6.9150306407196585, -1.5010913467717295, -0.8620147671516013, 3.4253782151811087, -1.0861512725567382, -2.394418890504598, 0.565866488278207, -0.9389187548185367, 0.012928107826576954, -0.10957757537484092], "ratio96": 1.7911731889735683, "ratio_final": 1.8080969842886498, "diag": {"ratio": 1.8080969842886498, "nonnormal": 0.8708825700618938, "eig_gaps": [0.02768309762312823, 0.28281770839722503, 0.3046736104705863], "discness": 0.029892918580847104, "peaks": 3, "deg": 3, "diamW": 9.943285777825906}, "class": "stall", "probe_best": 1.8362351175748777} +{"family": "random", "seed": 100012, "x": [0.8967272567614075, -0.15651867014372695, 0.5398891923780852, -0.4674217709579496, 0.8970601297828384, 1.3727485965576989, 0.03161356890473382, 0.4494922266254775, 2.8951017641286914, -0.061818973106778495, 2.0829344418107327, 0.26060728266378985, -7.745218066958293, -2.421821808767242, 7.127150618046123, -0.6845853931579513, 0.7077294632141942, 0.45945784020154345, -0.018871110219946495, 0.14342348765078133], "ratio96": 1.8738062500904267, "ratio_final": 1.8764272748780273, "diag": {"ratio": 1.8764272748780273, "nonnormal": 1.144760971694565, "eig_gaps": [0.1224074748229131, 0.3955241549234103, 0.4848171872777912], "discness": 0.053412524518147195, "peaks": 3, "deg": 3, "diamW": 3.86643213541663}, "class": "stall", "probe_best": 1.8869287186188324} +{"family": "random", "seed": 100013, "x": [0.13175643655830888, 1.0226764179756258, -0.5686335540539952, -0.5808675251261695, -0.605189211417031, -0.05690850740023082, 0.9736262291901138, 0.1641571196346642, 0.5387001937410778, 0.3499826370393766, 0.911355805525442, 0.1566964300237854, 0.43656832158320524, 2.13018831540922, 0.3107186725367111, 0.7753757671538739, -0.7388318416861706, 0.7961018570244298, 2.5016637565809687, -0.7578221352200192], "ratio96": 1.597413679610732, "ratio_final": 1.6031759311287956, "diag": {"ratio": 1.6031759311287956, "nonnormal": 0.8178727242284826, "eig_gaps": [0.22415051316697931, 0.5578369079720052, 0.7467641321001369], "discness": 0.22659498685072613, "peaks": 3, "deg": 3, "diamW": 2.343214266034889}, "class": "stall", "probe_best": 1.6079212713889623} +{"family": "random", "seed": 100014, "x": [1.0468110637672692, 0.8232206356930973, 0.18926094822752476, 0.18444533642676636, 0.10829603061002815, -0.028659306202212083, -1.742005772071236, 0.5086619267262648, -0.07487533795516776, -0.23908815039995523, 0.540040685575835, 1.9688058417241363, -0.18310285646294258, 0.3164219720692881, -1.145195388656929, -2.8733772794930275, 1.959198179791363, 1.3310585528106555, 0.16384897016027816, -0.22401381628602846], "ratio96": 1.9886263962978807, "ratio_final": 1.9885905206198318, "diag": {"ratio": 1.9885905206198318, "nonnormal": 0.645178401216383, "eig_gaps": [0.07449051439711116, 0.3494083111647348, 0.4141626082762303], "discness": 0.060569395857676095, "peaks": 4, "deg": 3, "diamW": 3.060348017051216}, "class": "near2-double", "probe_best": 1.9885905206198318} +{"family": "random", "seed": 100015, "x": [-2.1082139756864624, 0.23884001126931353, -0.4114913182764382, 3.1501249018710364, -2.051164793159149, -0.43532548139755367, -1.3123815616312808, 3.0748708867493164, 2.3158677784617687, 1.1114821504581536, -3.1646242526117634, -0.7588138137300793, 0.39845018850002556, 3.927162138568426, -1.6479803642655744, -2.173372165956893, 1.5159437798306712, -1.0780797617645228, 1.1990053965647085, 3.4196562781015762], "ratio96": 1.3006169812587736, "ratio_final": 1.3024986626469275, "diag": {"ratio": 1.3024986626469275, "nonnormal": 0.6141943083065777, "eig_gaps": [0.09924812678090553, 0.4942985997425767, 0.5783457402032008], "discness": 0.19077435839326132, "peaks": 2, "deg": 3, "diamW": 6.817005163037982}, "class": "stall", "probe_best": 1.405586236354821} +{"family": "random", "seed": 100016, "x": [0.8855717261374931, -0.986801248822396, -1.5510361006529625, 0.3715518529460575, -0.2780486767213678, -1.893455669923906, 0.2625838417742547, 1.9480854490017474, -0.499204578021452, -0.7720579174309954, -1.0063451362533977, 0.6261918697000821, -2.7597104808422155, -0.09493384257917951, -1.8523188499980514, 0.42961912707207095, -0.062899659703959, -0.0437858810585586, -1.1650734209784832, -0.23681707064583124], "ratio96": 1.5552965852362042, "ratio_final": 1.5670927509184347, "diag": {"ratio": 1.5670927509184347, "nonnormal": 0.6731105266698615, "eig_gaps": [0.38550914501288686, 0.6790128386025932, 0.7290418142088296], "discness": 0.1913532425812262, "peaks": 3, "deg": 3, "diamW": 3.826469196766727}, "class": "stall", "probe_best": 1.5744426953867874} +{"family": "random", "seed": 100017, "x": [1.0287461261539532, 1.2683852202957018, -0.7323843408820017, -0.5147354085025595, -0.22611277050726758, -0.9169059999715552, -0.7817194926779311, -0.15268296797625647, -0.015486163062248218, -0.013524136640631422, 1.020373671088426, -0.5406192907252994, 2.526359674484584, -1.2141298110432592, -0.2425924844006807, -1.355998069414206, 0.7335048292066964, 1.4702372148552274, -0.4344720437208043, 0.22593552765310443], "ratio96": 1.5900209688306248, "ratio_final": 1.6198671741374246, "diag": {"ratio": 1.6198671741374246, "nonnormal": 0.5418713695916001, "eig_gaps": [0.204527672555345, 0.7927837168440192, 0.7971306440680052], "discness": 0.2561837339125295, "peaks": 4, "deg": 3, "diamW": 3.16128115167602}, "class": "stall", "probe_best": 1.710932785433892} +{"family": "random", "seed": 100018, "x": [0.09410923068666305, -1.9233788381140888, -2.1206326917619025, 0.09899525629092926, -0.21772330590101086, 1.9926444110562977, 0.6436765457511338, 1.864426444249358, 1.838681079570482, -0.8916715747512178, -0.06476959643985636, -1.7104379142955608, 2.132619424098193, 1.5014730676408752, -1.0169609297919768, -0.3664483077649636, -1.4408405343754165, -2.0032800971151, 0.7650627789199858, 0.9799465095022653], "ratio96": 1.4601121954285377, "ratio_final": 1.4605379236497829, "diag": {"ratio": 1.4605379236497829, "nonnormal": 0.7475897396241189, "eig_gaps": [0.5223141785713838, 0.5835229016256583, 0.7643165229878228], "discness": 0.11619123055840902, "peaks": 3, "deg": 3, "diamW": 5.139780586079189}, "class": "stall", "probe_best": 1.5337681143186976} +{"family": "random", "seed": 200000, "x": [-0.20727378146166764, 0.3981344437327161, 0.4081820052050273, -0.41123356302179004, -0.0069812020865957745, -0.20839220714428153, 0.8402530183646023, -0.8238515450598278, -0.02145183310759179, -1.976025632772055, 0.013509012797120662, -0.7022535634021403, -1.6675684966086641, 1.2646483783964078, 5.248475503882508, 2.468447473716778, -0.09625274812217019, 1.0379814924673356, -0.09181850346029227, -0.14668164440740286], "ratio96": 1.8908738872478639, "ratio_final": 1.8910074964009838, "diag": {"ratio": 1.8910074964009838, "nonnormal": 1.3921095537372083, "eig_gaps": [0.17682283429182985, 0.24443302256585037, 0.38910390201581835], "discness": 0.029905105434007682, "peaks": 3, "deg": 3, "diamW": 2.6131586871093586}, "class": "stall", "probe_best": 1.8962988388107553} +{"family": "random", "seed": 200001, "x": [1.728664181558529, -1.2054036884764938, 1.7205880017655877, -0.8723556450198927, -1.3247033629840672, -0.40449332191913223, -1.024754762636727, -0.024259417381703056, 0.2684113252725627, -0.08860505434914652, 0.06802138842657543, -0.4517842015150463, -1.3015791014064089, 1.5929895359876634, -0.10062829265360064, 1.8054427059431677, -0.44014665863955266, 0.24617820370383708, 0.7808880296240163, -0.011255383495726214], "ratio96": 1.6892139251234144, "ratio_final": 1.6447805433383824, "diag": {"ratio": 1.6447805433383824, "nonnormal": 0.2355657977807862, "eig_gaps": [0.0910615262496102, 0.8421610333867753, 0.86283624043961], "discness": 1.1243324443745881, "peaks": 1, "deg": 3, "diamW": 3.6584709658843364}, "class": "stall", "probe_best": 1.6723726212583632} +{"family": "random", "seed": 200002, "x": [0.7036858667098222, 0.03465367302166064, -0.4441043076149801, -1.6405032742836951, -0.3205106030705511, -2.3065517442296946, 0.8263722882289373, -1.3364650321243459, -0.2974900088067053, 0.3475049541015105, 0.2561116329729205, -1.1581896161216365, 1.813390822686939, 0.45413512852034604, -0.2360362620846292, -2.7658374553876577, 0.9272145473532967, -0.34338163177986836, -0.036529281231942407, -0.9148559084221135], "ratio96": 1.7372067786545988, "ratio_final": 1.7372606665458985, "diag": {"ratio": 1.7372606665458985, "nonnormal": 0.4506736411402936, "eig_gaps": [0.20587203050440897, 0.6171305912322188, 0.7766121374117598], "discness": 0.8860817059819379, "peaks": 4, "deg": 3, "diamW": 3.2904839215408703}, "class": "stall", "probe_best": 1.7482183240358005} +{"family": "random", "seed": 200003, "x": [-1.3743214302535147, -0.5337599163233602, 1.240589860952416, -0.28500973107990013, 1.3149030853138064, -0.8524495610396476, -3.788085856465466, -1.1083464444316453, 0.7950867625766467, 1.4505444755524026, 2.0579208535857303, 3.0633606709243217, 0.8841071593243347, 0.8667409053480515, 2.052859144615721, -1.3649963107947616, -1.4624078818513562, -0.44160139749351257, -0.1303950925533412, -0.008141492571012272], "ratio96": 1.982701264964469, "ratio_final": 1.9826930963618628, "diag": {"ratio": 1.9826930963618628, "nonnormal": 0.7291400387856981, "eig_gaps": [0.08998973031586566, 0.4130413540587588, 0.4258295455270663], "discness": 0.04965228571313416, "peaks": 3, "deg": 3, "diamW": 6.359450744887105}, "class": "near2-double", "probe_best": 1.9826930963618628} +{"family": "random", "seed": 200004, "x": [-0.5819813106715443, 1.3147314709442504, 0.8562643129383087, 0.2422458716907742, -0.09755886589176387, 0.5535611226701924, 1.0579334382058905, 0.1881476703714531, -0.4150674472542538, 0.6290396666315244, 1.472498059657569, -0.67416433365892, -0.6308342626055181, 2.000648650167602, -0.9015354163563793, -1.0722545755085147, -0.25608669623326796, -0.41550373421042647, -0.02809745550027475, -1.3098205031623205], "ratio96": 1.6853439113507511, "ratio_final": 1.688628520465699, "diag": {"ratio": 1.688628520465699, "nonnormal": 0.7133592911991993, "eig_gaps": [0.33138163114402663, 0.3685133744861647, 0.65894547955512], "discness": 0.18859961988492482, "peaks": 4, "deg": 3, "diamW": 2.7226754979491634}, "class": "stall", "probe_best": 1.6903839053780803} +{"family": "random", "seed": 200005, "x": [-1.7763091973999783, 0.8930667031766439, -2.334031671829977, 2.3514070343183366, -0.9822769146387179, 0.49432987177660664, -0.7810617049881149, 2.136792587419132, -0.34727309162392683, 0.42095261288367514, -0.9598914761406885, 0.5271645160790852, 1.9547639183128176, 2.023682025123879, 0.1927026988072379, -5.420294787756656, 2.2758874965918183, -2.5403579112617356, 0.37424600128967933, 0.00851113006009735], "ratio96": 1.728879357216773, "ratio_final": 1.7400715415585755, "diag": {"ratio": 1.7400715415585755, "nonnormal": 0.4608897941007086, "eig_gaps": [0.28133419218022615, 0.4943703514327459, 0.7272836735359233], "discness": 97491640427.07242, "peaks": 3, "deg": 3, "diamW": 3.158257879250238}, "class": "stall", "probe_best": 1.7417003863162015} +{"family": "random", "seed": 200006, "x": [-1.1867442619669133, -0.18322713634091506, -0.17252403071202732, -0.12682795707977845, -0.4481802417413013, 0.13938432552596336, -0.600682909918804, -0.22064171820991674, 0.911529320654126, 0.4144319803608643, -2.0189309524311017, -0.09511774676553948, -0.08332757922696742, 1.5138934313302532, 3.39078944724512, 5.131137300616616, -0.32429200864747454, 0.8334407199240683, -0.1259297145123836, 0.08430703507708848], "ratio96": 1.9194946462574012, "ratio_final": 1.9320959989558406, "diag": {"ratio": 1.9320959989558406, "nonnormal": 1.0791410412136497, "eig_gaps": [0.1610758628515134, 0.33876112746063536, 0.4269611232353744], "discness": 0.005369616133957167, "peaks": 3, "deg": 3, "diamW": 2.379109244467514}, "class": "stall", "probe_best": 1.933408480768646} +{"family": "random", "seed": 200007, "x": [-1.3188147981527751, -0.8812639275998809, -0.47682886667044033, -0.7957519141715963, 0.9012693611816208, 0.4891760055060247, -1.5479060510493396, 1.3702302710517384, -0.9066277830379411, -0.5405935609681444, -1.6266283414428813, -0.12800885049922822, -0.5898128283670115, -1.333092769365011, 0.6031315501988863, 0.41625572682205586, 1.7472976723965745, -0.26321119439716933, 0.2918951296747627, -0.05612905245416744], "ratio96": 1.8334682426179096, "ratio_final": 1.8404060520565506, "diag": {"ratio": 1.8404060520565506, "nonnormal": 0.7543408155798433, "eig_gaps": [0.21664983758480963, 0.4823378359038694, 0.6678812978389246], "discness": 0.16762283477296416, "peaks": 4, "deg": 3, "diamW": 3.9063822863877036}, "class": "stall", "probe_best": 1.925482068475871} +{"family": "random", "seed": 200008, "x": [0.27446160385066654, 0.8904057668834517, -0.4717453437440977, 0.154914762331391, -0.2860223425017522, -0.9835358728376744, -1.1476704882262165, 2.7369158914672465, -0.14575978404939383, -0.44169591859480584, -2.487288116964902, -0.14665294296929457, 0.8380322331654317, -1.2305070370056042, 0.9064111750001056, 0.20346545209579048, 4.163104348176575, -0.5527002698213168, 0.16006754871661522, 0.37994579097642067], "ratio96": 1.9604209069900405, "ratio_final": 1.9606714033995576, "diag": {"ratio": 1.9606714033995576, "nonnormal": 0.7455096492967256, "eig_gaps": [0.24173312916681747, 0.2661324521182717, 0.45127496156685043], "discness": 0.03758914559683087, "peaks": 3, "deg": 3, "diamW": 4.3343089680623255}, "class": "stall", "probe_best": 1.972435106180648} +{"family": "random", "seed": 200009, "x": [-0.6812149662906526, 0.7982630795488297, -0.32595521020856955, 0.3092947730989428, 1.081417143146723, -0.02793177287718536, 1.6467791494637676, -0.9714703493902487, -0.14292649189095671, 0.17354338695474059, 0.5527816891396454, 1.5644243956102475, 1.7601709357596986, -1.131367287182092, -0.0021364634194132176, 2.0731740838356525, -2.603630044647645, -0.33291986498414916, -0.3753057709105121, 0.24639490358703703], "ratio96": 1.9623745914281074, "ratio_final": 1.9624579595306595, "diag": {"ratio": 1.9624579595306595, "nonnormal": 0.6905489223217623, "eig_gaps": [0.18639708612803516, 0.44631998355729, 0.6003490265760766], "discness": 0.11000790062073583, "peaks": 4, "deg": 3, "diamW": 3.242540575866021}, "class": "stall", "probe_best": 1.9656673171911776} +{"family": "random", "seed": 200010, "x": [1.0108937698855707, 1.2754605959847085, -1.4877926648331774, 0.8258449262181704, 0.4061811154268076, -2.8149775251594367, -3.5498989124627167, 8.617568031417054, 3.0640871147093156, 0.3646144030796036, -9.037152760847755, -4.05982137782857, 1.1106012120650721, 0.10329074478410809, -0.35299225678509516, -0.5791008259472236, 8.574878320960586, 1.786105826042435, 0.01277539625381129, -0.216120234741269], "ratio96": 1.9934017869517426, "ratio_final": 1.993412142555695, "diag": {"ratio": 1.993412142555695, "nonnormal": 0.7330638223975042, "eig_gaps": [0.17399610393520115, 0.2812641782691535, 0.28338234765448267], "discness": 0.011380715962464708, "peaks": 4, "deg": 3, "diamW": 14.591224927632792}, "class": "near2-other", "probe_best": 1.993412142555695} +{"family": "random", "seed": 200011, "x": [-0.6465181908533031, 1.2704531693633747, 0.568425549659316, -0.11720233352479711, -0.18391333674163968, 0.9562999476082259, 0.732571608246267, -1.3200729957553032, 0.4328730380444179, -0.3621423736765227, -1.7448159943668244, 0.1514445517180306, 1.3601186478751166, 1.0136675955773704, -3.32865050914726, 1.590611514684781, -0.8679640214174233, -1.3333252270030047, -0.40196095201907767, 0.5053116080229865], "ratio96": 1.9333906189423427, "ratio_final": 1.9347119896752438, "diag": {"ratio": 1.9347119896752438, "nonnormal": 0.6247412824654508, "eig_gaps": [0.18547740078019742, 0.43480543705997765, 0.6117537857494899], "discness": 0.14546964848867736, "peaks": 4, "deg": 3, "diamW": 3.014878487729709}, "class": "stall", "probe_best": 1.9590293631127655} +{"family": "random", "seed": 200012, "x": [-0.05461819400177076, -0.22824944904616407, 0.18326828997632094, 1.6584016657866982, -0.28491744978610434, -0.862597902152686, -1.128874132292126, 5.0246850090524955, -1.019229291316892, 0.1929738707271077, 1.9045701537244657, -4.446399221588061, 2.1310161988267287, -1.482630830684182, -0.08996511180125644, -0.8436753827518877, 2.645067652480142, 0.38626167951106216, 0.06994300533397627, 0.09572981749401205], "ratio96": 1.9763509872238738, "ratio_final": 1.9772965831464346, "diag": {"ratio": 1.9772965831464346, "nonnormal": 0.7311629885305081, "eig_gaps": [0.08876352533174069, 0.25011386147863823, 0.33725382488277456], "discness": 0.025772635467157683, "peaks": 3, "deg": 3, "diamW": 7.60289506306437}, "class": "near2-double", "probe_best": 1.9772965831464346} +{"family": "random", "seed": 300000, "x": [-0.6813971460903179, -1.2848840813851146, -1.339020672011971, 1.0098866639500517, 0.24676675870780712, 1.787344625904916, 1.1722941958586832, -0.4649873870203385, 1.3566799672867726, 0.32755287630465524, 1.3110922453490503, 0.3841055956892929, 2.740360126135154, -0.8018516067419499, -0.2167722315132442, -1.7584518949630745, 2.3042913404746823, 0.38515598678947843, -1.9213134422030715, 0.9175485779198191], "ratio96": 1.4891877278943595, "ratio_final": 1.4918182509573383, "diag": {"ratio": 1.4918182509573383, "nonnormal": 0.6799578250820644, "eig_gaps": [0.44343655822668104, 0.5993638280416301, 0.8058101413026433], "discness": 0.20887319596784268, "peaks": 3, "deg": 3, "diamW": 3.9827909964259254}, "class": "stall", "probe_best": 1.5047851461082322} +{"family": "random", "seed": 300001, "x": [1.0937842144431515, 0.83274592168023, 0.7293348557125535, 0.6888527270560572, 0.3847719209480996, -0.13623658618806356, -0.6464885414530894, 0.742791424140099, 0.2922793546260005, -0.46862271676625933, -0.3408720627303237, 0.010990409621225187, -0.5184655551958094, 2.0666411323257137, 0.8419492620226359, 1.0222945129604994, -1.8268681646465046, -1.3874322844060658, 0.15226597199832653, 0.5478260939497093], "ratio96": 1.622862989400532, "ratio_final": 1.6745373486800608, "diag": {"ratio": 1.6745373486800608, "nonnormal": 0.4818309828226202, "eig_gaps": [0.23126988882847985, 0.5277553091638165, 0.7086808519724832], "discness": 101182349291.00082, "peaks": 3, "deg": 3, "diamW": 1.6942423532266246}, "class": "stall", "probe_best": 1.7962924985821194} +{"family": "random", "seed": 300002, "x": [-1.4108468126436782, -0.8150076598169205, -0.8192461175160368, 0.5817137743986165, 1.6013165744567144, 2.4270080768161773, 8.244595481139303, -1.498946253181685, -4.206762601428409, 0.28715153722992176, 0.26656856232427706, -0.43158278179802134, 0.6996688302900804, 5.257847331916345, -4.235903091129684, 11.829835254447799, 0.4454818256010298, -0.14213806448108812, 0.038596871052335106, -0.005835696806673812], "ratio96": 1.6741084789315808, "ratio_final": 1.7273642851384579, "diag": {"ratio": 1.7273642851384579, "nonnormal": 1.280543382167043, "eig_gaps": [0.150064374713862, 0.3011208692167919, 0.4378088112892793], "discness": 0.032150545960073974, "peaks": 2, "deg": 3, "diamW": 10.107969472524063}, "class": "stall", "probe_best": 1.7287715856080978} +{"family": "random", "seed": 300003, "x": [0.03463492844153197, -0.025642005586560095, -0.18807761262324754, -0.020173804886019272, -0.24481903925963253, -0.061484985280982056, -0.20834696741826642, 0.27657325240557284, 0.5521993764710049, -1.1553441950862882, -0.19773289777498096, 0.14824189551161726, 0.10075134969284388, 0.0811146132469798, 2.7532404811640165, 0.18128867725305275, 0.2448502780840856, -0.15421183358074514, -0.03460527082251129, 0.0029766691142853454], "ratio96": 1.9884591281052564, "ratio_final": 1.988703121753976, "diag": {"ratio": 1.988703121753976, "nonnormal": 1.3551469540154806, "eig_gaps": [0.051177553663092186, 0.1624422149863272, 0.20543613272658412], "discness": 0.007716772043572913, "peaks": 3, "deg": 3, "diamW": 1.3714394413725468}, "class": "near2-double", "probe_best": 1.988703121753976} +{"family": "random", "seed": 300004, "x": [-0.9829717052081945, -0.9446515711907095, -1.8880326164896515, -1.0061706850062677, -0.560049296396607, -0.10880949480801098, 2.0737302016117325, -0.379141897390643, -0.5652476527927236, -0.45991092606560524, -1.6832404689727336, -0.004035576245346575, 2.158819201115916, -1.1343427865993334, 0.7102886056988673, 2.6344218975092897, 1.5571749809987425, 1.4444478337895412, 0.27646930883014437, -0.1920440744574744], "ratio96": 1.6505309955660237, "ratio_final": 1.650451247896331, "diag": {"ratio": 1.650451247896331, "nonnormal": 0.5390381807968079, "eig_gaps": [0.2846968444155609, 0.29398575862865184, 0.5030008966811926], "discness": 7.110133204962462, "peaks": 3, "deg": 3, "diamW": 3.1863693570846494}, "class": "stall", "probe_best": 1.651067615488703} +{"family": "random", "seed": 300005, "x": [1.880185524829294, 1.6315069139863851, -0.8761970175452758, 0.03673729447734848, 0.7453622794804844, -0.11257074978164959, -0.4503130269419089, 0.5375450621622844, 1.0439837128594207, 2.296561355857119, -5.835727307908487, -0.9491255864587901, -0.9103875214934811, -0.4806505588259842, 4.184626895137727, 3.0151362401971507, 0.008048640410676334, 0.022446292802832654, -0.01687589183643868, -0.001468933915153878], "ratio96": 1.9526163224866457, "ratio_final": 1.952715171666236, "diag": {"ratio": 1.952715171666236, "nonnormal": 1.241542540416192, "eig_gaps": [0.2350444409525572, 0.3003373410010333, 0.45964538184527565], "discness": 0.026787099398347676, "peaks": 3, "deg": 3, "diamW": 6.92813107293926}, "class": "stall", "probe_best": 1.9950856402668642} +{"family": "random", "seed": 300006, "x": [-1.6469368732262222, 0.017983387101466387, 1.0883866771580464, 1.0038260359043667, -0.9318371381350692, -0.9497270411623036, -1.617074771904253, 2.144542185973137, 1.0122689854710174, -1.7617243545165073, 2.4777657458099207, 0.8726333400607929, -2.86577216370038, 0.06473508949316581, -0.06278925185681883, -2.0488445187142386, 0.9424676261145213, -2.7062482271504873, -0.02882380623853216, -0.19466501040099315], "ratio96": 1.9237036356739134, "ratio_final": 1.9236629681671338, "diag": {"ratio": 1.9236629681671338, "nonnormal": 0.6840887789911051, "eig_gaps": [0.22479283718237023, 0.5250166976333388, 0.5431897594008801], "discness": 0.09072079220852611, "peaks": 3, "deg": 3, "diamW": 5.352744112903607}, "class": "stall", "probe_best": 1.951443948157988} +{"family": "random", "seed": 300007, "x": [0.4971175184934504, 0.4737179460029658, -0.30044941323948327, 0.3550634400349575, 0.4487368841452438, 0.14220531948325865, 0.8421873088538354, 0.6182786364931401, 1.6630744615235187, -1.223861506862476, 0.015479854857265538, 1.1853495692624807, 0.5339634153876449, -0.8532352450747527, 0.9789967773123469, 2.147770522042337, -0.00018578050133811162, 0.15077589927740576, -0.012889017272980474, -0.021584965580598817], "ratio96": 1.982780607293167, "ratio_final": 1.9828824168910333, "diag": {"ratio": 1.9828824168910333, "nonnormal": 1.3143341743805745, "eig_gaps": [0.12229512466217568, 0.2843020159811236, 0.2943428997443752], "discness": 0.02314266615310511, "peaks": 3, "deg": 3, "diamW": 2.7394741967662823}, "class": "near2-double", "probe_best": 1.9828824168910333} +{"family": "random", "seed": 300008, "x": [0.8740602120477032, 0.8839503856982609, -0.4922414817391889, -0.35777244525857294, 1.0641312999525254, -0.04895812896610098, -2.416052660890498, -1.4918764488308967, 0.986380727716115, -0.4169070755869627, -2.486460234137772, 1.630953610961164, -0.4494118951814293, 0.3216806216814504, 1.9038235244523956, 1.2988671884278014, -1.9607316571016462, -0.7724398191943074, -0.020266733094105926, -0.020433466240021764], "ratio96": 1.9649935391645037, "ratio_final": 1.9669303506095994, "diag": {"ratio": 1.9669303506095994, "nonnormal": 0.6812694842320294, "eig_gaps": [0.2039461072331789, 0.33989390199414116, 0.39549062881060454], "discness": 0.03382987400832846, "peaks": 3, "deg": 3, "diamW": 4.66826373980073}, "class": "stall", "probe_best": 1.984836378585894} +{"family": "random", "seed": 300009, "x": [-0.5363711204223878, -1.2134891963489811, -0.06498936736616445, 1.9229074430760829, -1.3395410644063666, -0.29523171254496783, -0.9785563171384722, -0.6399011644277162, 0.15822049478349975, -1.0156554852927617, 0.9340264581550215, -0.09488945227501336, -2.2730628568454394, 5.5729490795437755, -0.04203608558068267, 0.6141043255679818, -1.4863996268615292, -0.6589378116620452, 3.2082851807202006, 0.04387327515047054], "ratio96": 1.4916871554286917, "ratio_final": 1.4998936239348117, "diag": {"ratio": 1.4998936239348117, "nonnormal": 0.5665072787725145, "eig_gaps": [0.3262590732960073, 0.6841680419867868, 0.848206986769885], "discness": 0.3224291972854775, "peaks": 3, "deg": 3, "diamW": 3.7392073127210765}, "class": "stall", "probe_best": 1.5027724476461812} +{"family": "random", "seed": 300010, "x": [-0.2599246785450231, 1.3149740839683353, -1.1122434291108116, -0.9652775008580277, 0.7585854684727411, -0.8046856673932159, 2.8303851850281756, 2.4217040619040393, 0.8832971625625394, 1.802435780928307, 0.2236838474592382, 3.204851649062456, -0.6360705075435915, -0.5291770114292194, -0.3603608539326466, 1.1328286103143705, 1.052445780723155, 1.437605805778707, -0.14850086448122732, 0.1246753806901515], "ratio96": 1.9676406517698055, "ratio_final": 1.9676430185679914, "diag": {"ratio": 1.9676430185679914, "nonnormal": 0.7846604045974668, "eig_gaps": [0.3248485370233035, 0.4068439004463389, 0.4211465885996627], "discness": 0.012912215592314613, "peaks": 4, "deg": 3, "diamW": 5.7802595040744364}, "class": "below2-max", "probe_best": 1.968086907630799} +{"family": "random", "seed": 300011, "x": [-1.389248586348946, -0.13289810290739293, 1.1259073580296164, 1.6340402987500575, 0.5422646815419448, 0.019688078001725978, -2.2170539497408566, -5.435430030777681, 1.5650094536448655, 2.5713151623550616, -0.12867636033236013, -0.13503751044322987, 1.8420704154228116, -0.2849771189960923, 1.9082078385384773, 4.620052997280258, -0.0906345250929824, 0.0973893110400646, -0.019189793011511233, 0.0005615459095831504], "ratio96": 1.9663925996580267, "ratio_final": 1.9663806978337794, "diag": {"ratio": 1.9663806978337794, "nonnormal": 1.3586690607004073, "eig_gaps": [0.236209373828989, 0.26660759158420216, 0.4229564661825396], "discness": 0.03870303113435805, "peaks": 3, "deg": 3, "diamW": 7.267350878881991}, "class": "stall", "probe_best": 1.9825586667435318} +{"family": "random", "seed": 300012, "x": [-0.7926056305264266, -1.1505961102883981, 0.8033814174411891, 0.6697344040139148, -0.19745229162106237, 0.37733143556113546, 1.642121186803954, -1.3190803291797766, -0.23407237980591789, -1.467067121156143, -2.057396802478521, 0.6169618204153873, -0.9584044188894634, -1.3017904302860237, 0.22899673760498174, -0.03442512217194016, 1.857674633273652, -2.0070773222034948, 0.025836432004742088, -0.23413729814538886], "ratio96": 1.9462546977126907, "ratio_final": 1.946212916612244, "diag": {"ratio": 1.946212916612244, "nonnormal": 0.7397553311876851, "eig_gaps": [0.2530887130000983, 0.39801675864627034, 0.5876275578237576], "discness": 0.0753038626186094, "peaks": 3, "deg": 3, "diamW": 4.119793755239455}, "class": "stall", "probe_best": 1.9485521206923506} diff --git a/problems/crouzeix/data/classify.log b/problems/crouzeix/data/classify.log new file mode 100644 index 0000000..c4bc897 --- /dev/null +++ b/problems/crouzeix/data/classify.log @@ -0,0 +1,26 @@ +286 endpoints + probed 50/286 + probed 100/286 + probed 150/286 + probed 200/286 + probed 250/286 + +class counts: {'stall': 175, 'near2-double': 48, 'near2-other': 44, 'one-exact': 13, 'below2-max': 4, 'near2-triple': 2} + companion {'stall': 30, 'near2-double': 1, 'one-exact': 1, 'below2-max': 2} + jordan2 {'near2-other': 12, 'stall': 7, 'near2-double': 14, 'near2-triple': 1} + jordan3 {'near2-double': 11, 'near2-other': 15, 'stall': 9, 'near2-triple': 1} + nearnormal {'stall': 25, 'one-exact': 8, 'near2-double': 1} + peaks {'stall': 22, 'near2-other': 8, 'near2-double': 4} + random {'stall': 64, 'one-exact': 3, 'near2-other': 6, 'near2-double': 7, 'below2-max': 1} + real {'stall': 18, 'near2-double': 10, 'below2-max': 1, 'near2-other': 3, 'one-exact': 1} + +ratio distribution of non-stall endpoints: + 1.0: 13 + 1.8: 1 + 2.0: 97 + +4 probe-surviving endpoints strictly between 1 and 1.97: + random seed=300010 ratio=1.96764 gaps=[0.325, 0.407, 0.421] disc=0.013 peaks=4 nonnormal=0.785 + companion seed=604002 ratio=1.96026 gaps=[0.163, 0.451, 0.591] disc=0.107 peaks=4 nonnormal=0.688 + real seed=406004 ratio=1.95884 gaps=[0.191, 0.286, 0.413] disc=0.040 peaks=3 nonnormal=0.774 + companion seed=504010 ratio=1.78117 gaps=[0.301, 0.482, 0.75] disc=0.320 peaks=4 nonnormal=0.647 diff --git a/problems/crouzeix/data/deep_probe.log b/problems/crouzeix/data/deep_probe.log new file mode 100644 index 0000000..7abb39a --- /dev/null +++ b/problems/crouzeix/data/deep_probe.log @@ -0,0 +1,4 @@ +real seed=406004 ratio96=1.95884 ratio512=1.95884 best_escape=1.96343 STALL +companion seed=504010 ratio96=1.78117 ratio512=1.78117 best_escape=1.78652 STALL +companion seed=604002 ratio96=1.96026 ratio512=1.96026 best_escape=1.96476 STALL +random seed=300010 ratio96=1.96764 ratio512=1.96764 best_escape=1.97300 STALL diff --git a/problems/crouzeix/explore/census.py b/problems/crouzeix/explore/census.py new file mode 100644 index 0000000..8372903 --- /dev/null +++ b/problems/crouzeix/explore/census.py @@ -0,0 +1,505 @@ +#!/usr/bin/env python3 +"""Local-maxima census for the Crouzeix ratio, n = 3, deg p <= 3. + +Float-only EXPLORATION tool (per the verification contract, nothing here is a +claim). It maximizes the Crouzeix ratio + + f(A, p) = ||p(A)|| / max_{z in W(A)} |p(z)| + +over 3x3 upper-triangular complex A (Schur form is WLOG: the ratio is +unitary-similarity invariant) and polynomials p of degree <= 3, by +Nelder-Mead restarts plus a compass-search polish, from several *structured* +start families. Endpoints are re-evaluated at high boundary resolution and +classified by invariant features. Certification of representative endpoints +is done separately by the tier-0 harness (see certify_endpoints.py). + +Pure standard library. Usage: + census.py --pilot a few starts, sanity output + census.py --run N --seed S --out FILE N starts per family, JSON lines +""" + +import argparse +import cmath +import json +import math +import random +import sys +import time + +# ---------------------------------------------------------------------- +# 3x3 complex linear algebra (lists of lists of python complex) +# ---------------------------------------------------------------------- + +def mat_mul(a, b): + return [[sum(a[i][k] * b[k][j] for k in range(3)) for j in range(3)] + for i in range(3)] + + +def mat_add(a, b): + return [[a[i][j] + b[i][j] for j in range(3)] for i in range(3)] + + +def mat_scale(s, a): + return [[s * a[i][j] for j in range(3)] for i in range(3)] + + +def adjoint(a): + return [[a[j][i].conjugate() for j in range(3)] for i in range(3)] + + +def eye(): + return [[1.0 + 0j if i == j else 0j for j in range(3)] for i in range(3)] + + +def fro(a): + return math.sqrt(sum(abs(a[i][j]) ** 2 for i in range(3) for j in range(3))) + + +def herm_part(a): + aa = adjoint(a) + return [[(a[i][j] + aa[i][j]) / 2 for j in range(3)] for i in range(3)] + + +def poly_mat(coeffs, a): + out = [[0j] * 3 for _ in range(3)] + power = eye() + for c in coeffs: + out = mat_add(out, mat_scale(c, power)) + power = mat_mul(power, a) + return out + + +# ---------------------------------------------------------------------- +# Hermitian 3x3 eigenvalues (trig cubic) and top eigenvector +# ---------------------------------------------------------------------- + +def herm_eigs(h): + """Real eigenvalues of Hermitian 3x3, ascending.""" + # charpoly: l^3 - c2 l^2 + c1 l - c0 + c2 = (h[0][0] + h[1][1] + h[2][2]).real + c1 = 0.0 + for (i, j) in ((0, 1), (0, 2), (1, 2)): + c1 += (h[i][i] * h[j][j] - h[i][j] * h[j][i]).real + c0 = det3(h).real + # shift l = m + c2/3 -> m^3 + p m + q + p = c1 - c2 * c2 / 3.0 + q = -2.0 * c2 ** 3 / 27.0 + c2 * c1 / 3.0 - c0 + # m^3 + p m + q = 0, all roots real for Hermitian + p = min(p, 0.0) + if p > -1e-300: + r = -q ** (1.0 / 3.0) if q >= 0 else (-q) ** (1.0 / 3.0) + roots = [r, r, r] if q >= 0 else [r, r, r] + # p ~ 0: triple root of shifted cubic + m = (-q) ** (1.0 / 3.0) if q <= 0 else -(q ** (1.0 / 3.0)) + roots = [m, m, m] + else: + a = math.sqrt(-p / 3.0) + arg = 3.0 * q / (2.0 * p * a) + arg = max(-1.0, min(1.0, arg)) + phi = math.acos(arg) / 3.0 + roots = [2 * a * math.cos(phi - 2 * math.pi * k / 3.0) + for k in range(3)] + roots = sorted(r + c2 / 3.0 for r in roots) + return roots + + +def det3(m): + return (m[0][0] * (m[1][1] * m[2][2] - m[1][2] * m[2][1]) + - m[0][1] * (m[1][0] * m[2][2] - m[1][2] * m[2][0]) + + m[0][2] * (m[1][0] * m[2][1] - m[1][1] * m[2][0])) + + +def cross(u, v): + """Formal (unconjugated) cross product; lies in ker M when rows of M + span rank <= 2 and u, v are rows of M.""" + return [u[1] * v[2] - u[2] * v[1], + u[2] * v[0] - u[0] * v[2], + u[0] * v[1] - u[1] * v[0]] + + +def top_eigvec(h, lam): + """Eigenvector of Hermitian 3x3 h for eigenvalue lam (float tolerant).""" + m = [[h[i][j] - (lam if i == j else 0.0) for j in range(3)] + for i in range(3)] + best, bn = None, -1.0 + rows = m + for (i, j) in ((0, 1), (0, 2), (1, 2)): + v = cross(rows[i], rows[j]) + n = sum(abs(x) ** 2 for x in v) + if n > bn: + best, bn = v, n + if bn > 1e-24: + n = math.sqrt(bn) + return [x / n for x in best] + # rank <= 1 (multiplicity 2): any vector plain-orthogonal to the + # largest row is in the kernel + r = max(rows, key=lambda row: sum(abs(x) ** 2 for x in row)) + rn = sum(abs(x) ** 2 for x in r) + if rn < 1e-24: + return [1.0, 0.0, 0.0] + # two independent candidates, pick the longer + cands = [[-r[1], r[0], 0j], [-r[2], 0j, r[0]], [0j, -r[2], r[1]]] + v = max(cands, key=lambda w: sum(abs(x) ** 2 for x in w)) + n = math.sqrt(sum(abs(x) ** 2 for x in v)) + return [x / n for x in v] + + +def quad_form(a, v): + """v* A v for unit v.""" + out = 0j + for i in range(3): + for j in range(3): + out += v[i].conjugate() * a[i][j] * v[j] + return out + + +# ---------------------------------------------------------------------- +# the ratio (float, boundary-sampled) +# ---------------------------------------------------------------------- + +def spec_norm(m): + g = mat_mul(adjoint(m), m) + return math.sqrt(max(0.0, herm_eigs(g)[2])) + + +def peval(coeffs, z): + out = 0j + for c in reversed(coeffs): + out = out * z + c + return out + + +def boundary_points(a, ntheta): + pts = [] + for k in range(ntheta): + th = 2 * math.pi * k / ntheta + e = cmath.exp(-1j * th) + h = herm_part(mat_scale(e, a)) + lam = herm_eigs(h)[2] + v = top_eigvec(h, lam) + pts.append(quad_form(a, v)) + return pts + + +def crouzeix_ratio_float(a, coeffs, ntheta=96, pts=None): + num = spec_norm(poly_mat(coeffs, a)) + if pts is None: + pts = boundary_points(a, ntheta) + den = max(abs(peval(coeffs, z)) for z in pts) + if den < 1e-14: + return 0.0 + return num / den + + +# ---------------------------------------------------------------------- +# parametrization: x in R^20 -> (A upper triangular, p) +# ---------------------------------------------------------------------- + +def unpack(x): + a = [[0j] * 3 for _ in range(3)] + a[0][0] = complex(x[0], x[1]) + a[1][1] = complex(x[2], x[3]) + a[2][2] = complex(x[4], x[5]) + a[0][1] = complex(x[6], x[7]) + a[0][2] = complex(x[8], x[9]) + a[1][2] = complex(x[10], x[11]) + p = [complex(x[12 + 2 * k], x[13 + 2 * k]) for k in range(4)] + return a, p + + +def pack(a, p): + x = [a[0][0].real, a[0][0].imag, a[1][1].real, a[1][1].imag, + a[2][2].real, a[2][2].imag, a[0][1].real, a[0][1].imag, + a[0][2].real, a[0][2].imag, a[1][2].real, a[1][2].imag] + for c in p: + x += [c.real, c.imag] + return x + + +def objective(x, ntheta=96): + a, p = unpack(x) + na = fro(a) + if na < 1e-9 or na > 50.0: + return -1.0 + np_ = math.sqrt(sum(abs(c) ** 2 for c in p)) + if np_ < 1e-9: + return -1.0 + p = [c / np_ for c in p] # pure gauge: ratio is p-scale invariant + return crouzeix_ratio_float(a, p, ntheta) + + +# ---------------------------------------------------------------------- +# optimizers: Nelder-Mead (maximize) + compass polish +# ---------------------------------------------------------------------- + +def nelder_mead(f, x0, step=0.25, maxit=2500, ftol=1e-10): + n = len(x0) + simplex = [list(x0)] + for i in range(n): + y = list(x0) + y[i] += step + simplex.append(y) + vals = [f(s) for s in simplex] + for _ in range(maxit): + order = sorted(range(n + 1), key=lambda i: -vals[i]) + simplex = [simplex[i] for i in order] + vals = [vals[i] for i in order] + if vals[0] - vals[-1] < ftol: + break + cen = [sum(s[i] for s in simplex[:-1]) / n for i in range(n)] + worst = simplex[-1] + refl = [2 * cen[i] - worst[i] for i in range(n)] + fr = f(refl) + if fr > vals[0]: + exp = [3 * cen[i] - 2 * worst[i] for i in range(n)] + fe = f(exp) + if fe > fr: + simplex[-1], vals[-1] = exp, fe + else: + simplex[-1], vals[-1] = refl, fr + elif fr > vals[-2]: + simplex[-1], vals[-1] = refl, fr + else: + con = [(cen[i] + worst[i]) / 2 for i in range(n)] + fc = f(con) + if fc > vals[-1]: + simplex[-1], vals[-1] = con, fc + else: + best = simplex[0] + for i in range(1, n + 1): + simplex[i] = [(simplex[i][j] + best[j]) / 2 + for j in range(n)] + vals[i] = f(simplex[i]) + order = sorted(range(n + 1), key=lambda i: -vals[i]) + return simplex[order[0]], vals[order[0]] + + +def compass(f, x0, v0, step=0.02, minstep=1e-6, maxit=4000): + x, v = list(x0), v0 + n = len(x) + it = 0 + while step > minstep and it < maxit: + improved = False + for i in range(n): + for s in (step, -step): + it += 1 + y = list(x) + y[i] += s + fy = f(y) + if fy > v + 1e-13: + x, v = y, fy + improved = True + break + if not improved: + step /= 2.0 + return x, v + + +# ---------------------------------------------------------------------- +# start families +# ---------------------------------------------------------------------- + +def rnd_c(rng, s=1.0): + return complex(rng.gauss(0, s), rng.gauss(0, s)) + + +def schur_of(a): + """No exact Schur needed: our families are built upper triangular.""" + return a + + +def start(family, rng): + if family == "random": + # baseline: dense-equivalent random upper triangular + random p + a = [[rnd_c(rng) if j >= i else 0j for j in range(3)] for i in range(3)] + p = [rnd_c(rng) for _ in range(4)] + elif family == "jordan3": + # perturbed nilpotent Jordan block, p near z^2 (known 3x3 extremal) + eps = 0.3 + a = [[eps * rnd_c(rng) if j >= i else 0j for j in range(3)] + for i in range(3)] + a[0][1] += 1.0 + a[1][2] += 1.0 + p = [eps * rnd_c(rng) for _ in range(4)] + p[2] += 1.0 + elif family == "jordan2": + # perturbed J2 (+) [c], p near z (the embedded 2x2 extremal) + eps = 0.3 + a = [[eps * rnd_c(rng) if j >= i else 0j for j in range(3)] + for i in range(3)] + a[0][1] += 1.0 + a[2][2] += 0.3 * rnd_c(rng) + p = [eps * rnd_c(rng) for _ in range(4)] + p[1] += 1.0 + elif family == "nearnormal": + # normal (diagonal) + small strictly-upper perturbation; p a cubic + # with roots near the spectrum (peaks |p| at several boundary points) + d = [rnd_c(rng) for _ in range(3)] + eps = 0.15 + a = [[0j] * 3 for _ in range(3)] + for i in range(3): + a[i][i] = d[i] + a[0][1], a[0][2], a[1][2] = (eps * rnd_c(rng), eps * rnd_c(rng), + eps * rnd_c(rng)) + # p with roots at perturbed midpoints of the spectrum + r1 = (d[0] + d[1]) / 2 + 0.2 * rnd_c(rng) + r2 = (d[1] + d[2]) / 2 + 0.2 * rnd_c(rng) + r3 = (d[0] + d[2]) / 2 + 0.2 * rnd_c(rng) + p = poly_from_roots([r1, r2, r3]) + elif family == "companion": + # companion matrix of a random cubic (upper Hessenberg made upper + # triangular is wrong; instead use its transpose-free triangular + # surrogate: random triangular with the same spectrum), p lacunary z^3 + roots = [rnd_c(rng) for _ in range(3)] + a = [[0j] * 3 for _ in range(3)] + for i in range(3): + a[i][i] = roots[i] + # strong coupling, as in a companion form + a[0][1], a[1][2], a[0][2] = 1.0 + 0j, 1.0 + 0j, rnd_c(rng, 0.5) + p = [0j, 0j, 0j, 1.0 + 0j] + p = [p[k] + 0.1 * rnd_c(rng) for k in range(4)] + elif family == "peaks": + # elliptic-ish W(A); p = product of linear factors just outside the + # boundary at spread angles -> multiple near-coincident |p| peaks + a = [[0j] * 3 for _ in range(3)] + a[0][0], a[1][1], a[2][2] = -1.0 + 0j, 0j, 1.0 + 0j + a[0][1] = 0.8 + 0.4 * rng.random() + a[1][2] = 0.8 + 0.4 * rng.random() + a[0][2] = 0.3 * rnd_c(rng) + pts = boundary_points(a, 24) + zs = rng.sample(pts, 3) + # push each root outward by 10-40% + c = sum(pts, 0j) / len(pts) + roots = [c + (z - c) * (1.1 + 0.3 * rng.random()) for z in zs] + p = poly_from_roots(roots) + elif family == "real": + # real upper triangular, real polynomial + a = [[complex(rng.gauss(0, 1), 0) if j >= i else 0j + for j in range(3)] for i in range(3)] + p = [complex(rng.gauss(0, 1), 0) for _ in range(4)] + else: + raise ValueError(family) + return pack(a, p) + + +def poly_from_roots(roots): + p = [1.0 + 0j] + for r in roots: + new = [0j] * (len(p) + 1) + for i, c in enumerate(p): + new[i + 1] += c + new[i] -= r * c + p = new + return p + + +# ---------------------------------------------------------------------- +# endpoint diagnostics (invariant features) +# ---------------------------------------------------------------------- + +def diagnostics(x): + a, p = unpack(x) + np_ = math.sqrt(sum(abs(c) ** 2 for c in p)) + p = [c / np_ for c in p] + pts = boundary_points(a, 512) + r = crouzeix_ratio_float(a, p, pts=pts) + na = fro(a) + # non-normality (unitary-similarity invariant) + comm = mat_add(mat_mul(adjoint(a), a), mat_scale(-1, mat_mul(a, adjoint(a)))) + nonnormal = fro(comm) / (na * na) + # eigenvalues of A: diagonal (A is triangular throughout the search) + evs = [a[0][0], a[1][1], a[2][2]] + scale = max(abs(z1 - z2) for z1 in pts for z2 in pts) # diam W(A) + gaps = sorted(abs(evs[i] - evs[j]) / max(scale, 1e-12) + for i in range(3) for j in range(i + 1, 3)) + # disc-ness of W(A): fit support function h(th) ~ Re(e^{-i th} c) + R + n = len(pts) + hs, cs, ss = [], [], [] + for k in range(n): + th = 2 * math.pi * k / n + e = cmath.exp(-1j * th) + h = herm_part(mat_scale(e, a)) + hs.append(herm_eigs(h)[2]) + cs.append(math.cos(th)) + ss.append(math.sin(th)) + R = sum(hs) / n + cr = 2 * sum(h * c for h, c in zip(hs, cs)) / n + ci = 2 * sum(h * s for h, s in zip(hs, ss)) / n + resid = math.sqrt(sum((h - (R + cr * c + ci * s)) ** 2 + for h, c, s in zip(hs, cs, ss)) / n) + discness = resid / max(R - math.hypot(cr, ci) + 1e-12, 1e-12) + # |p| peak structure on the boundary + vals = [abs(peval(p, z)) for z in pts] + vmax = max(vals) + peaks = 0 + for k in range(n): + if vals[k] >= 0.99 * vmax and vals[k] >= vals[k - 1] \ + and vals[k] >= vals[(k + 1) % n]: + peaks += 1 + # degree of p and root positions relative to W(A) centre/diam + deg = max((k for k, c in enumerate(p) if abs(c) > 1e-7), default=0) + return { + "ratio": r, + "nonnormal": nonnormal, + "eig_gaps": gaps, + "discness": discness, + "peaks": peaks, + "deg": deg, + "diamW": scale, + } + + +# ---------------------------------------------------------------------- +# driver +# ---------------------------------------------------------------------- + +FAMILIES = ["random", "jordan3", "jordan2", "nearnormal", "companion", + "peaks", "real"] + + +def one_start(family, seed): + rng = random.Random(seed) + x0 = start(family, rng) + f96 = lambda x: objective(x, 96) + x1, v1 = nelder_mead(f96, x0, step=0.2, maxit=1500) + x2, v2 = nelder_mead(f96, x1, step=0.05, maxit=1000) + f192 = lambda x: objective(x, 192) + x3, v3 = compass(f192, x2, f192(x2), step=0.01, maxit=1500) + d = diagnostics(x3) + return {"family": family, "seed": seed, "x": x3, + "ratio96": v2, "ratio_final": d["ratio"], "diag": d} + + +def main(): + ap = argparse.ArgumentParser() + ap.add_argument("--pilot", action="store_true") + ap.add_argument("--run", type=int, default=0, metavar="N", + help="starts per family") + ap.add_argument("--seed", type=int, default=0) + ap.add_argument("--families", type=str, default=",".join(FAMILIES)) + ap.add_argument("--out", type=str, default="") + args = ap.parse_args() + + fams = args.families.split(",") + if args.pilot: + for fam in fams: + t0 = time.time() + rec = one_start(fam, 1000 + hash(fam) % 1000) + dt = time.time() - t0 + print(f"{fam:11s} ratio={rec['ratio_final']:.6f} " + f"({dt:.1f}s) diag={ {k: (round(v,4) if isinstance(v,float) else v) for k,v in rec['diag'].items()} }") + return + + out = open(args.out, "w") if args.out else sys.stdout + # interleave families so an early cutoff stays unbiased across them + for i in range(args.run): + for fam in fams: + rec = one_start(fam, args.seed * 100000 + FAMILIES.index(fam) * 1000 + i) + out.write(json.dumps(rec) + "\n") + out.flush() + if args.out: + out.close() + + +if __name__ == "__main__": + main() diff --git a/problems/crouzeix/explore/certify_endpoints.py b/problems/crouzeix/explore/certify_endpoints.py new file mode 100644 index 0000000..71b9bc2 --- /dev/null +++ b/problems/crouzeix/explore/certify_endpoints.py @@ -0,0 +1,111 @@ +#!/usr/bin/env python3 +"""Certified enclosures for census endpoints, via the tier-0 harness. + +Each endpoint (A upper triangular 3x3 complex, p of degree <= 3) is rounded +entrywise to denominator DEN, and harness/crouzeix/ratio.py computes a +certified enclosure [ratio_lo, ratio_hi] of the Crouzeix ratio *of the rounded +point*, together with the exact-rational refutation test +||p(A)||^2 > 4 (max_{W} |p|)^2. The claim certified is therefore about the +rounded rational configurations, which lie within 2^-1/DEN of the float +endpoints entrywise; the float values are exploration only. + +Top endpoints additionally go through verify_ratio.verify, the independent +route (Faddeev-LeVerrier + Sturm), per the verification contract. + +Usage: + certify_endpoints.py classified.jsonl --den 256 --out certified.jsonl \ + [--top-verify 8] [--limit N] +""" + +import argparse +import json +import sys +import time +from fractions import Fraction as Q + +sys.path.insert(0, "/home/user/work-crouzeix/harness/crouzeix") +import ratio as R +import verify_ratio as V +sys.path.insert(0, "/home/user/work-crouzeix/explore") +from census import unpack + + +def to_rational(x, den): + a, p = unpack(x) + A = [[R.Cx(Q(round(a[i][j].real * den), den), + Q(round(a[i][j].imag * den), den)) for j in range(3)] + for i in range(3)] + # normalize p as the census did (unit norm) BEFORE rounding, so the + # rounded coefficients are O(1) + import math + nrm = math.sqrt(sum(abs(c) ** 2 for c in p)) + P = [R.Cx(Q(round(c.real / nrm * den), den), + Q(round(c.imag / nrm * den), den)) for c in p] + return A, P + + +def main(): + ap = argparse.ArgumentParser() + ap.add_argument("infile") + ap.add_argument("--den", type=int, default=256) + ap.add_argument("--out", required=True) + ap.add_argument("--top-verify", type=int, default=0) + ap.add_argument("--limit", type=int, default=0) + ap.add_argument("--tol", type=int, default=6, help="tol = 10^-this") + ap.add_argument("--dirs", type=int, default=3) + args = ap.parse_args() + + recs = [json.loads(line) for line in open(args.infile)] + if args.limit: + recs = recs[:args.limit] + tol = Q(1, 10 ** args.tol) + + t_all = time.time() + out = open(args.out, "w") + n_ok = n_ref = 0 + for i, rec in enumerate(recs): + A, P = to_rational(rec["x"], args.den) + t0 = time.time() + r = R.crouzeix_ratio(A, P, dir_bound=args.dirs, tol=tol) + dt = time.time() - t0 + if r["degenerate"]: + row = {"seed": rec["seed"], "family": rec["family"], + "class": rec.get("class"), "degenerate": True} + else: + if r["refutes"]: + n_ref += 1 + # floor the exact rational at 12 decimals so the stored decimal + # is still a certified lower bound + lo_scaled = int(r["ratio_lo"] * 10 ** 12) # exact floor + lo_dec = f"{lo_scaled // 10**12}.{lo_scaled % 10**12:012d}" + row = {"seed": rec["seed"], "family": rec["family"], + "class": rec.get("class"), + "float_ratio": rec["diag"]["ratio"], + "ratio_lo": lo_dec, + "ratio_hi_f": float(r["ratio_hi"]), + "refutes": r["refutes"], "den": args.den, + "dirs": args.dirs, "tol_exp": args.tol, "sec": round(dt, 2)} + n_ok += 1 + out.write(json.dumps(row) + "\n") + out.flush() + if (i + 1) % 25 == 0: + print(f" {i+1}/{len(recs)} certified, " + f"{time.time()-t_all:.0f}s elapsed", file=sys.stderr) + out.close() + print(f"certified {n_ok} points, refutations: {n_ref}, " + f"total {time.time()-t_all:.0f}s") + + if args.top_verify: + recs.sort(key=lambda r: -r["diag"]["ratio"]) + print(f"\nindependent verification (verify_ratio.py) of the top " + f"{args.top_verify} endpoints by float ratio:") + for rec in recs[:args.top_verify]: + A, P = to_rational(rec["x"], args.den) + print(f"[{rec['family']} seed={rec['seed']} " + f"float={rec['diag']['ratio']:.5f}]") + ok = V.verify(A, P, dir_bound=args.dirs, tol=tol) + print(" verify:", "AGREE" if ok else "MISMATCH") + + +if __name__ == "__main__": + main() diff --git a/problems/crouzeix/explore/classify.py b/problems/crouzeix/explore/classify.py new file mode 100644 index 0000000..da16060 --- /dev/null +++ b/problems/crouzeix/explore/classify.py @@ -0,0 +1,120 @@ +#!/usr/bin/env python3 +"""Classify census endpoints and probe whether sub-2 endpoints are genuine +local maxima or optimizer stalls. + +Escape probe: at each endpoint with final ratio < 1.97, try NP random +perturbation directions at several scales; record the best ratio found. An +endpoint from which random perturbation finds ascent is a STALL, not a local +maximum, and is excluded from any 'local maxima all look like X' statement. + +Float-only exploration tool; claims are certified separately. +""" + +import json +import math +import random +import sys + +sys.path.insert(0, "/home/user/work-crouzeix/explore") +from census import objective, unpack, diagnostics, nelder_mead + +SCALES = (1e-3, 3e-3, 1e-2, 3e-2) +NPROBE = 60 # directions per scale + + +def escape_probe(x, v, rng, ntheta=96): + """Random perturbations at several scales; if none ascend, one full + Nelder-Mead restart (a fresh simplex sees ridge directions random + probes miss).""" + best = v + n = len(x) + for s in SCALES: + for _ in range(NPROBE): + y = [xi + rng.gauss(0, s) for xi in x] + fy = objective(y, ntheta) + if fy > best: + best = fy + if best <= v + 5e-4: + _, vnm = nelder_mead(lambda z: objective(z, 96), x, step=0.08, + maxit=600) + best = max(best, vnm) + return best + + +def classify(rec, escaped): + d = rec["diag"] + r = d["ratio"] + if escaped: + return "stall" + if r > 1.97: + # near-2: split by eigenvalue structure + g = d["eig_gaps"] # sorted pairwise |ev_i - ev_j| / diam W + if g[2] < 0.15: + return "near2-triple" # all eigenvalues coincide: J3-like + if g[0] < 0.15: + return "near2-double" # two coincide: J2 (+) [c]-like + return "near2-other" + if abs(r - 1.0) < 1e-6: + return "one-exact" + if r < 1.97: + return "below2-max" # survived the probe below 1.97 + return "other" + + +def main(): + files = sys.argv[1:] + recs = [] + for fn in files: + with open(fn) as f: + for line in f: + recs.append(json.loads(line)) + print(f"{len(recs)} endpoints") + rng = random.Random(12345) + out = [] + for i, rec in enumerate(recs): + v = rec["diag"]["ratio"] + escaped = False + best = v + if v < 1.97: + best = escape_probe(rec["x"], objective(rec["x"], 96), rng) + escaped = best > v + 5e-4 + cls = classify(rec, escaped) + rec["class"] = cls + rec["probe_best"] = best + out.append(rec) + if (i + 1) % 50 == 0: + print(f" probed {i+1}/{len(recs)}", file=sys.stderr) + with open("/home/user/work-crouzeix/data/classified.jsonl", "w") as f: + for rec in out: + f.write(json.dumps(rec) + "\n") + + # summary + from collections import Counter, defaultdict + byc = Counter(r["class"] for r in out) + print("\nclass counts:", dict(byc)) + byfc = defaultdict(Counter) + for r in out: + byfc[r["family"]][r["class"]] += 1 + for fam in sorted(byfc): + print(f" {fam:11s} {dict(byfc[fam])}") + print("\nratio distribution of non-stall endpoints:") + hist = Counter() + for r in out: + if r["class"] != "stall": + hist[round(r["diag"]["ratio"], 1)] += 1 + for k in sorted(hist): + print(f" {k:4.1f}: {hist[k]}") + # the interesting ones: survived probing, not near 2, not exactly 1 + odd = [r for r in out if r["class"] == "below2-max"] + odd.sort(key=lambda r: -r["diag"]["ratio"]) + print(f"\n{len(odd)} probe-surviving endpoints strictly between 1 and 1.97:") + for r in odd[:20]: + d = r["diag"] + print(f" {r['family']:11s} seed={r['seed']} ratio={d['ratio']:.5f} " + f"gaps={[round(g,3) for g in d['eig_gaps']]} " + f"disc={d['discness']:.3f} peaks={d['peaks']} " + f"nonnormal={d['nonnormal']:.3f}") + + +if __name__ == "__main__": + main() diff --git a/problems/crouzeix/explore/deep_probe.py b/problems/crouzeix/explore/deep_probe.py new file mode 100644 index 0000000..0a2cedf --- /dev/null +++ b/problems/crouzeix/explore/deep_probe.py @@ -0,0 +1,32 @@ +#!/usr/bin/env python3 +"""Aggressive escape attempt on selected endpoints: high-resolution objective +(512 boundary samples), multiple NM restarts with jitter, 1200 random probes. +An endpoint is a STALL if anything ascends by > 1e-3.""" +import json, random, sys +sys.path.insert(0, "/home/user/work-crouzeix/explore") +from census import objective, nelder_mead, compass + +def deep(x, v0, rng): + f = lambda z: objective(z, 512) + v = f(x) + best = v + for s in (1e-3, 3e-3, 1e-2, 3e-2, 1e-1): + for _ in range(240): + y = [xi + rng.gauss(0, s) for xi in x] + best = max(best, objective(y, 128)) + for k in range(3): + x0 = [xi + rng.gauss(0, 0.02) for xi in x] + _, vnm = nelder_mead(lambda z: objective(z, 128), x0, step=0.1, maxit=1200) + best = max(best, vnm) + xc, vc = compass(f, x, v, step=0.005, maxit=1200) + best = max(best, vc) + return v, best + +rng = random.Random(999) +targets = [json.loads(l) for l in open("/home/user/work-crouzeix/data/classified.jsonl") + if json.loads(l)["class"] == "below2-max"] +for r in targets: + v512, best = deep(r["x"], r["diag"]["ratio"], rng) + verdict = "STALL" if best > v512 + 1e-3 else "SURVIVES" + print(f"{r['family']} seed={r['seed']} ratio96={r['diag']['ratio']:.5f} " + f"ratio512={v512:.5f} best_escape={best:.5f} {verdict}", flush=True) diff --git a/problems/crouzeix/explore/measure_cost.py b/problems/crouzeix/explore/measure_cost.py new file mode 100644 index 0000000..30e8157 --- /dev/null +++ b/problems/crouzeix/explore/measure_cost.py @@ -0,0 +1,18 @@ +import sys, os, time +sys.path.insert(0, "/home/user/work-crouzeix/harness/crouzeix") +import ratio +from ratio import Cx, Q + +def mat(entries, den): + return [[Cx(Q(a, den), Q(b, den)) for (a, b) in row] for row in entries] + +# dense 3x3, denominator 32 (what we'd round float endpoints to) +A32 = mat([[(7,-3),(19,5),(-11,2)],[(3,8),(-5,-13),(17,1)],[(2,-9),(6,4),(-1,11)]], 32) +p32 = [Cx(Q(5,32),Q(-3,32)), Cx(Q(-7,32),Q(2,32)), Cx(Q(11,32),Q(1,32)), Cx(Q(9,32),Q(-4,32))] + +for label, den, A, p in [("den32", 32, A32, p32)]: + for tol_exp, db in [(6, 2), (9, 3)]: + t0 = time.time() + r = ratio.crouzeix_ratio(A, p, dir_bound=db, tol=Q(1, 10**tol_exp)) + dt = time.time() - t0 + print(f"{label} tol=1e-{tol_exp} dirs={db}: {dt:.2f}s ratio=[{float(r['ratio_lo']):.6f},{float(r['ratio_hi']):.6f}] refutes={r['refutes']}") diff --git a/problems/crouzeix/explore/survivors_detail.py b/problems/crouzeix/explore/survivors_detail.py new file mode 100644 index 0000000..6dc1fa7 --- /dev/null +++ b/problems/crouzeix/explore/survivors_detail.py @@ -0,0 +1,104 @@ +#!/usr/bin/env python3 +"""Geometric detail for probe-surviving census endpoints (float, exploration). + +For each endpoint of a chosen class: eigenvalues of A and roots of p placed +relative to W(A) (inside / on the boundary / outside, with signed margin as a +fraction of diam W), |p|-peak angles on the boundary, and the active set: +which boundary points attain max |p| and whether ||p(A)|| is attained with a +multiple top singular value of p(A). +""" + +import cmath +import json +import math +import sys + +sys.path.insert(0, "/home/user/work-crouzeix/explore") +from census import (unpack, boundary_points, peval, herm_part, mat_scale, + herm_eigs, poly_mat, mat_mul, adjoint, fro) + + +def roots_dk(coeffs): + """Durand-Kerner for a complex polynomial, ascending coefficients.""" + cs = list(coeffs) + while cs and abs(cs[-1]) < 1e-10: + cs.pop() + n = len(cs) - 1 + if n < 1: + return [] + lead = cs[-1] + cs = [c / lead for c in cs] + rs = [complex(0.4, 0.9) ** k for k in range(n)] + for _ in range(300): + new = [] + for i, r in enumerate(rs): + num = 0j + v = 0j + for c in reversed(cs): + v = v * r + c + den = 1.0 + 0j + for j, s in enumerate(rs): + if j != i: + den *= (r - s) + new.append(r - v / den if abs(den) > 1e-30 else r) + shift = max(abs(a - b) for a, b in zip(rs, new)) + rs = new + if shift < 1e-13: + break + return rs + + +def margin(z, a, ntheta=256): + """max_theta [Re(e^{-i th} z) - h(th)] / diamW: negative inside W(A).""" + worst = -1e18 + for k in range(ntheta): + th = 2 * math.pi * k / ntheta + e = cmath.exp(-1j * th) + h = herm_part(mat_scale(e, a)) + lam = herm_eigs(h)[2] + worst = max(worst, (e * z).real - lam) + return worst + + +def detail(rec): + a, p = unpack(rec["x"]) + np_ = math.sqrt(sum(abs(c) ** 2 for c in p)) + p = [c / np_ for c in p] + pts = boundary_points(a, 512) + diam = max(abs(u - v) for u in pts for v in pts) + evs = [a[0][0], a[1][1], a[2][2]] + rts = roots_dk(p) + ev_m = [margin(z, a) / diam for z in evs] + rt_m = [margin(z, a) / diam for z in rts] + vals = [abs(peval(p, z)) for z in pts] + vmax = max(vals) + peak_idx = [k for k in range(len(pts)) + if vals[k] >= 0.999 * vmax + and vals[k] >= vals[k - 1] and vals[k] >= vals[(k + 1) % len(pts)]] + peak_angles = [round(360.0 * k / len(pts), 1) for k in peak_idx] + # top singular-value multiplicity of p(A) + g = mat_mul(adjoint(poly_mat(p, a)), poly_mat(p, a)) + s = herm_eigs(g) + sv_gap = (s[2] - s[1]) / max(s[2], 1e-15) + return { + "family": rec["family"], "seed": rec["seed"], + "ratio": rec["diag"]["ratio"], + "deg_p": len(rts), + "eig_margins": [round(m, 4) for m in ev_m], + "root_margins": [round(m, 4) for m in sorted(rt_m)], + "n_peaks": len(peak_idx), "peak_angles": peak_angles, + "sv_gap": round(sv_gap, 6), + } + + +def main(): + cls = sys.argv[2] if len(sys.argv) > 2 else "below2-max" + recs = [json.loads(line) for line in open(sys.argv[1])] + recs = [r for r in recs if r.get("class") == cls] + recs.sort(key=lambda r: -r["diag"]["ratio"]) + for r in recs: + print(json.dumps(detail(r))) + + +if __name__ == "__main__": + main() diff --git a/problems/crouzeix/prior-art.json b/problems/crouzeix/prior-art.json index 64d032c..c276e31 100644 --- a/problems/crouzeix/prior-art.json +++ b/problems/crouzeix/prior-art.json @@ -1,6 +1,32 @@ { "problem": "crouzeix", - "route_status": "NOT_ATTEMPTED", - "route_summary": "Onboarded, never worked. No prior art, so blind and informed mode are currently equivalent for this problem.", - "attempts": [] -} + "route_status": "EVIDENCE", + "route_summary": "First attempt (blind, 2026-08-04): a certified 286-start local-maxima census at n=3, deg<=3 found no unknown extremal structure; skeptic pass pending. The published intermediate GO maxima are untested by this design and are the standing gap.", + "attempts": [ + { + "id": "001", + "file": "attempts/001-blind-dim3-local-maxima-census.md", + "date": "2026-08-04", + "mode": "blind", + "status": "EVIDENCE", + "mechanism": [ + "optimization-landscape-census", + "nonsmooth-maximization", + "certified-enclosure", + "basin-classification" + ], + "one_line": "286-start local-maxima census of the Crouzeix ratio at n=3, deg p<=3 from 7 structured start families: every probe-surviving local maximum is known structure (13 ice-cream-cone points at ratio 1, 97 Jordan-type endpoints near ratio 2); all 4 intermediate candidates were optimizer stalls; all 286 endpoints certified below 2, top 8 verified by the independent harness route.", + "leak_terms": [ + "escape-probe stall rate 61%", + "near2-other class", + "deep-probe stall verdict", + "286-start census n3 deg3", + "ice-cream-cone rediscovery at ratio 1" + ], + "range": "dimension 3, polynomial degree <= 3; 286 optimization starts (families: random 81, perturbed-J3 36, perturbed-J2+c 34, near-normal 34, companion-like 34, multi-peak 34, real 33); certified enclosures at all 286 endpoints rounded to denominator 256 (tol 1e-6, 32 half-planes), 0 refutations, max certified lower bound 1.969725", + "gaps": [ + "published intermediate local maxima (ratios ~1.185, ~1.433 at this slice, per Greenbaum-Overton Calcolo 2021) were never reached by this design; their basins are untested here and the census says nothing about them" + ] + } + ] +} \ No newline at end of file From f5c4cef6d852662746e8884bbf39368191c96216 Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 11:31:07 +0000 Subject: [PATCH 13/14] crouzeix 002: skeptic review of 001 - census stands, five corrections Independent re-certification of 30/286 endpoints with a structurally different certifier (Sylvester-minor bisection, Bernstein segment bounds, own direction set) - all consistent, anchors to the last digit, denominator discretization confirmed conservative. All four intermediate-candidate stall verdicts re-confirmed with a corrected equal-sample-count probe protocol (001's 128-vs-512 comparison could manufacture ascent; verdicts survive at 512-vs-512). Corrections: near-2 class is 94 not 97 (partition now exact); index one_line's 'all certified below 2' overstated - the sound statement is 0 certified above 2; Calcolo 2021 / arXiv:2105.14176 is Overton alone; ascent figures precise; 2/13 ratio-1 maxima have near-flat multipeaks. Blind label accurate for prior-art absence; task framing was orchestrator-assigned (caveat recorded). Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- .../attempts/002-skeptic-review-of-001.md | 305 ++++++++ problems/crouzeix/data/skeptic_probe.jsonl | 9 + problems/crouzeix/data/skeptic_probe.log | 9 + problems/crouzeix/data/skeptic_recert.jsonl | 30 + problems/crouzeix/data/skeptic_recert.log | 32 + problems/crouzeix/explore/skeptic_probe.py | 229 ++++++ problems/crouzeix/explore/skeptic_recert.py | 690 ++++++++++++++++++ problems/crouzeix/prior-art.json | 40 +- 8 files changed, 1341 insertions(+), 3 deletions(-) create mode 100644 problems/crouzeix/attempts/002-skeptic-review-of-001.md create mode 100644 problems/crouzeix/data/skeptic_probe.jsonl create mode 100644 problems/crouzeix/data/skeptic_probe.log create mode 100644 problems/crouzeix/data/skeptic_recert.jsonl create mode 100644 problems/crouzeix/data/skeptic_recert.log create mode 100644 problems/crouzeix/explore/skeptic_probe.py create mode 100644 problems/crouzeix/explore/skeptic_recert.py diff --git a/problems/crouzeix/attempts/002-skeptic-review-of-001.md b/problems/crouzeix/attempts/002-skeptic-review-of-001.md new file mode 100644 index 0000000..c0d004e --- /dev/null +++ b/problems/crouzeix/attempts/002-skeptic-review-of-001.md @@ -0,0 +1,305 @@ +# 002 — Skeptic review of 001 (dim-3 local-maxima census): adversarial verification + +- **Problem:** Crouzeix's conjecture, `problems/crouzeix/PROBLEM.md` +- **Date:** 2026-08-04 +- **Mode:** informed +- **Type:** adversarial verification of `001-blind-dim3-local-maxima-census.md` + (default stance: refute). Certificates re-verified with an independent + implementation sharing no algorithm with the tier-0 harness; stall verdicts + re-probed with a different scheme, RNG, and consistent resolution; citations + checked against the papers. +- **Outcome in one line:** 001's EVIDENCE stands — 30/30 stratified + re-certifications consistent, both anchors reproduced to the digit, all 4 + stall verdicts confirmed, both attacked ratio-1 maxima survive — but the + record miscounts its near-2 class (94, not 97), its index entry overstates + "all 286 endpoints certified below 2" (the certificates prove *0 certified + above 2*; 182/286 enclosure upper ends exceed 2), and the Calcolo 2021 paper + it leans on is by Overton alone, not Greenbaum–Overton. +- **Tools:** `explore/skeptic_recert.py` (independent certifier: Sylvester + minors by cofactor expansion + exact Rayleigh lower bounds + integer-isqrt + root bounds + rounded-trig rational directions + Bernstein-coefficient + segment bounds; no hull-ordering code; deterministic; ~2.8 s/endpoint), + `explore/skeptic_probe.py` (sparse-perturbation escape probe, RNG 20260804, + 512-vs-512-sample confirmation, second float evaluator via complex Jacobi + rotations); checkpoints `data/skeptic_recert.jsonl`, `data/skeptic_probe.jsonl`, + logs alongside. Reproduce with: + `python3 explore/skeptic_recert.py --selftest | --anchors | --sample` and + `python3 explore/skeptic_probe.py stalls 300010 604002 406004 504010`, + `... maxima 403004 603009`, `... stallcheck 400000 403000 405000`. +- **Sources:** 001 and its data/tools; `harness/crouzeix/ratio.py` / + `verify_ratio.py` (read to audit soundness direction and understand data + formats — the re-certifier transcribes nothing from them); + arXiv:2105.14176 (Overton, Calcolo 2021) via the ar5iv machine rendering + [T]; Greenbaum–Overton preprint optimization-online 5703 (2016), PDF + fetched and text-extracted. + +## Claims attacked + +1. **Certificate validity** — the 286 den-256 rational endpoint enclosures + (tol 1e-6, 32 half-planes), the two calibration anchors, the 1.969725 + record lower bound, and the 1.974 certification ceiling at the exact 2×2 + Jordan pair. +2. **Direction of the W(A) discretization** — does the 32-half-plane outer + polygon inflate or deflate the denominator, and does the certified + inequality survive it? +3. **Classification arithmetic** — the 13/97/175/4 accounting against + `classified.jsonl` (13+97+175+4 = 289 ≠ 286), the 61% stall rate, and the + per-family start counts against the raw census files. +4. **The stall verdicts** — the 4 deep-probed intermediate candidates + (1.96764, 1.96026, 1.95884, 1.78117): does ascent really exist? The deep + probe compared 128-sample escape values to a 512-sample baseline — a + mismatch whose sign can manufacture apparent ascent. +5. **The ratio-1 maxima** — probe two of the 13 hard (ascent found there + would refute the census's only "genuine local maximum" class), and check + the claimed structure on all 13. +6. **Citations and transcriptions** — the ice-cream-cone attribution, the + intermediate values ~1.433/~1.185 ([L]-transcribed in 001), the "~half a + million heavy-tailed starts", and the claimed design differences from + Greenbaum–Overton. +7. **Mode and scope** — the blind label, tier-1 contamination, and an + overclaim audit of the Outcome section and the prior-art index entry. + +## Refutations found + +### R1. The near-2 class has 94 members, not 97 + +`classified.jsonl`: near2-double 48 + near2-other 44 + near2-triple 2 = **94** +(and 175 + 94 + 13 + 4 = 286, as it must). 001's own table says 48/44/2, but +its prose bullet says "The 97 near-2 endpoints", and the index `one_line` +repeats "97 Jordan-type endpoints near ratio 2". Corrected statement: **94 +near-2 endpoints** (48 double-coalescence, 44 separated-eigenvalue, 2 +triple-coalescence). The impossible sum 13+97+175+4 = 289 is exactly the +trace of this error. + +### R2. "All 286 endpoints certified below 2" (index entry) is not what was proved + +The certificates establish (i) a certified lower bound below 2 at every +endpoint and (ii) failure of the exact refutation test `n2_lo > 4·outer2` — +i.e. **no endpoint is certified above 2**. They do *not* certify any endpoint +*below* 2: in `certified.jsonl`, **182 of 286** enclosure upper ends +(`ratio_hi_f`) are ≥ 2 (up to 2.38), as expected for near-extremal points +enclosed with a 32-gon outer polygon. The attempt record's own Outcome text +states it correctly ("0 refutations; the largest certified lower bound … is +1.969725"); the `one_line` in `prior-art.json` overstates it. Corrected +statement: *all 286 endpoints certified — 0 refutations (none certified +above 2), max certified lower bound 1.969725*. + +### R3. Calcolo 2021 / arXiv:2105.14176 is by Overton alone + +001's References list "A. Greenbaum, M. L. Overton, *Local minimizers of the +Crouzeix ratio: a nonsmooth optimization case study*, Calcolo 58 (2021)", and +the prose attributes that paper's findings to "Greenbaum-Overton" throughout. +The paper is single-authored: **M. L. Overton** (verified against +arXiv:2105.14176). The 2018 LAA paper is correctly Greenbaum–Overton. Every +substantive claim 001 takes from the 2021 paper checks out (S6); only the +attribution is wrong. + +### R4. Minor: the deep-probe ascent figures, and a resolution mismatch in the probe protocol + +`deep_probe.log` shows ascents +0.00459, +0.00535, +0.00450, +0.00536 — i.e. +**0.0045–0.0054**, not "0.005-0.006" as the record says. More substantively, +`deep_probe.py` compares escape candidates evaluated at **128** boundary +samples against a baseline at **512**; coarser boundary sampling +under-estimates max_{∂W}|p| and so over-estimates the ratio, which can +manufacture ascent. The verdicts survive anyway (S4: consistent-resolution +re-probing finds genuine ascent at all four), but the protocol flaw is real, +and the same mismatch exists in `classify.py` (probe at 96 vs baseline at +512). Quantified: only 20/175 stalls show any self-inflation +(`objective(x, 96) > ratio512 + 5e-4`), and in all 20 the probe found ascent +beyond the point's own 96-sample value — so the 61% stall figure is robust; +future probe protocols should still compare like with like. + +### R5. Minor: "single |p|-peak" is not clean on 2 of the 13 ratio-1 endpoints + +Re-running the peak diagnostic at 512 samples: 11/13 endpoints show one peak; +seeds 500009 and 100003 show ~10 near-coincident peaks at the 0.999·max +threshold (|p| is near-flat along a boundary arc). In all 13 the *dominant* +peak sits at the boundary eigenvalue, so the configuration identification +stands; the sentence "the single |p|-peak at it" is an overstatement for +those two. + +### R6. Framing: the census design was assigned, not blind-convergent + +The blind label is accurate for what the worker read (tier-0 copy plus GO +papers via web). But the design brief — this exact slice (n=3, deg ≤ 3), +cost-measurement-before-sizing, and "state how the design differs from GO +before running it" — is the lab's own tier-1 queue item (STATUS.md queue +item 10, derived from `PRIOR-ART.md`'s editorial view), delivered through +the task assignment. No crouzeix tier-1 *findings* existed before 001, so +the results are uncontaminated; but the *choice of approach* must not be +counted as independent blind evidence about what an unprompted agent would +try. Nothing in the record's text presupposes non-public knowledge; this is +a caveat on how the mode field should be read, not a mislabel. + +## Claims that survive (and what was thrown at them) + +### S1. The certificates — independent re-implementation, 30/30 consistent + +`skeptic_recert.py` was written from the PROBLEM.md verification contract +with disjoint algorithms end to end: eigenvalue upper bounds by bisection on +**Sylvester's criterion** (leading principal minors by cofactor expansion — +not LDL, not Sturm/Faddeev–LeVerrier), lower bounds by **exact Rayleigh +quotients** at rationalized power-iteration vectors, square-root bounds by +**integer `math.isqrt` scaling** (no bisection), W(A) outer polygons from +**40 rounded-trig rational directions** with support offsets rounded *up* +(vs the harness's 32 Gaussian-integer coprime pairs), and max|p|² bounded by +**Bernstein coefficients over all pairwise vertex segments** — no +convex-hull-ordering code exists in it, the one place a subtle geometry bug +could hide. Its selftest brackets the literature values (2 for (J2,z), √2 +for (J3,z), 1 for a normal matrix). + +Stratified sample of 30/286 (all 4 below2-max, 5 of 13 one-exact, 13 near-2 +including the record holder, 8 stalls; `data/skeptic_recert.jsonl`): **all +30 consistent** — 001's certified lower bound never exceeds my certified +upper bound, my enclosure always overlaps or sits inside their slack, the +refutation test agrees (false) at every point, and every one of my lower +bounds is < 2. Highlights: record holder seed 406009: their lo 1.969725 vs +my enclosure [1.988100, 1.995295] — their bound is valid and conservative; +the four killed intermediates re-certify at [1.9395, 1.9519], +[1.9411, 1.9508], [1.9419, 1.9567], [1.7479, 1.7751]; the five sampled +ratio-1 endpoints re-certify at [1.000000, 1.000000] to 12 digits. Anchors: +001's `(J3, z²) → [1.980557335, 2.25]` and `(J2⊕[1/4], z) → [1.990266652, +2.0]` reproduce **to the last printed digit** (their pipeline, dir_bound 5, +tol 1e-9), and my independent certifier at 80 directions gives +[1.996454, 2.0] and [1.998226, 2.0] — both contain the true value 2, both +consistent with theirs. The certification ceiling reproduces too: their +pipeline at tol 1e-6 / 32 half-planes gives lo = 1.97417 at the exact +(J2, z) pair, so 1.969725 at rounded census endpoints is indeed "as close to +2 as the 32-gon slack allows". + +### S2. The discretization direction is sound + +Code audit of `ratio.py`: the outer polygon uses the *upper* ends of the +per-direction eigenvalue enclosures, the vertex-filter slack only *enlarges* +the polygon, and the interval-Horner edge bound is an upper bound — so the +denominator is **over**-estimated, hence `ratio_lo` and the refutation test +are conservative: the safe direction for "no refutation" claims (an +under-estimated denominator would have been fatal, per the contract, and +was the thing this review most expected to find). Empirically, on the +semi-analytic control A = diag(1, i, −1), p = z (true max_{W}|p| = 1, at the +eigenvalue hull): their `outer_max2` = 1.00000024 ≥ 1, `inner_max2` = 1.0 +≤ 1, enclosure [0.99999940, 1.0] brackets the true ratio 1, no refutation. +The upper enclosure end correctly uses certified-inner points only. No +direction error found. + +### S3. The counts — reconciled exactly + +Raw census files: 79+83+79 (interleaved i4/i5/i6) + 19+13+13 (truncated +w1/w2/w3, all family "random", confirming the "family-biased first launch" +note) = 286. Per-family: random 81, jordan3 36, jordan2 34, nearnormal 34, +companion 34, peaks 34, real 33 — matches the record and the index `range`. +Class partition: 175 stall / 48+44+2 near-2 / 13 one-exact / 4 below2-max += 286 (a true partition; the only error is R1's prose "97"). 175/286 = +61.2% ✓. `certified.jsonl`: 286 rows, 0 degenerate, 0 refutes, max +`ratio_lo` = 1.969724783 at seed 406009 ✓; max stall lower bound 1.9389 +("up to 1.94" ✓). `certify.log`: 286 certified in 214 s ✓; top-8 +independent verification all "AGREE", zero MISMATCH lines ✓. + +### S4. The four stall verdicts — confirmed under a fair protocol + +`skeptic_probe.py`: different scheme (900 sparse coordinate-subset Gaussian +perturbations, log-uniform scales 10^-3.5 to 10^-0.7, 3 jittered NM +restarts, fine compass), different RNG (20260804), and **every candidate +confirmed at 512-vs-512 samples** before counting, then cross-checked with +a second float evaluator built on complex Jacobi rotations (agrees with the +census evaluator to 1e-15 on controls). All four killed intermediates +escape genuinely: +0.0040 (seed 300010), +0.0010 (604002), +0.0100 +(406004), +0.0219 (504010) — the two largest ascents *exceed* what 001's +deep probe found, so the stall verdicts are if anything understated. Three +randomly spot-checked ordinary stalls also escape at consistent resolution +(+0.0526, +0.2337, +0.0059). No hidden intermediate local maximum surfaced. + +### S5. The ratio-1 maxima — survive attack, structure verified + +Seeds 403004 and 603009 were probed with the full S4 arsenal: **zero +ascent** (best512 − v512 = +0.000000 for both). All 13 one-exact endpoints +re-checked structurally at 512 samples: in every case one eigenvalue sits +on ∂W(A) (margin −0.0 relative to diam W) and the dominant |p|-peak is +exactly at it, matching the published characterization of the ratio-1 local +minimizers of the reciprocal — A ≈ U diag(λ, B)U* with a dominant outside +scalar block (GO 2018, Theorem 2 and the surrounding discussion), for which +"W(A) has a vertex at λ, and often has the appearance of an ice cream cone" +(Overton 2021). The rediscovery-with-citation framing is correct (mod R3's +authorship fix), and the den-256 certificates pin all five sampled ratio-1 +endpoints to [1.000000, 1.000000]. + +### S6. The transcribed GO facts — all check out against the papers + +From the ar5iv rendering of arXiv:2105.14176 [T] and the text-extracted +2016 GO preprint: the intermediate locally minimal reciprocals include +**0.698 at (n=3, m=3)** → ratio 1.4327 ≈ "~1.433" ✓ and **0.844 at (n=2, +m=3) and (n=3, m=3)** → 1.1848 ≈ "~1.185" ✓; initialization is heavy-tailed +(entries x·e^{αx²}, α = 2, x normal), 10,000 starts per configuration, +**nearly 500,000 runs** in total → "~half a million heavy-tailed random +starts" ✓; GO 2018 uses **Chebfun** for bd W(A) plus **BFGS and Gradient +Sampling** from normally distributed random starts ✓. So 001's four claimed +design differences (fixed slice; structured start families; derivative-free +NM + compass with escape probing; certified endpoints) are all real +differences from the published methodology. + +### S7. Scope and labels — no overclaim in the record body + +The Outcome opens with `EVIDENCE` scoped by dimension, degree, and search +design, and explicitly disclaims any statement about the conjecture; the +"every surviving local maximum sits at already-known structure" verdict is +scoped to this design in the same paragraph and honestly hedged for the +near2-other subclass ("consistent with … coalescence only in the limit" — +44 of the 94 near-2 points have *not* collapsed their eigenvalues, and the +record says so rather than claiming they are Jordan pairs). The SPECULATION +label sits inline on the {1,2}-only basin claim, as required. The declared +gap (GO's intermediate maxima never reached; their basins untested) is +accurate and is the honest headline limitation. The index `gaps` entry +matches. Only the index `one_line` fails this audit (R1, R2). + +## Verdict + +| # | Claim | Verdict | +|---|-------|---------| +| a | 286 certified enclosures; anchors; 1.969725 record; 1.974 ceiling | **CONFIRMED** (30/30 stratified sample by independent implementation; anchors to the digit) | +| b | Discretization direction sound for "no refutation" | **CONFIRMED** (code audit + semi-analytic normal control) | +| c | Counts 13/97/175/4 and per-family | **CORRECTED** — near-2 is 94 (48/44/2); everything else reconciles exactly | +| d | 4 intermediate candidates are stalls | **CONFIRMED** (fair-protocol re-probe, ascents up to +0.022), protocol flaw noted and quantified harmless | +| e | Ratio-1 ice-cream-cone rediscovery; GO transcriptions | **CONFIRMED**, except **CORRECTED** authorship (Calcolo 2021 is Overton solo); "single peak" soft on 2/13 | +| f | Blind mode; design independence from GO | **CONFIRMED** as labelled, with the design-was-assigned caveat (R6) | +| g | Scope honesty; index fidelity | Record body **CONFIRMED**; index `one_line` **CORRECTED** (R1, R2) | + +**Net assessment: 001's EVIDENCE stands.** The certificates are valid and +conservative, the landscape accounting is real, and the two headline +structures (ratio-1 cones, near-2 Jordan-type climbs) are what the record +says they are. The corrections are bookkeeping and attribution, not +substance — but two of them (94 vs 97, "certified below 2") live in the +index line future agents read first, which is exactly where errors compound. + +## Residual risk + +- The re-certification covered 30 endpoints + 2 anchors of 286; the + remaining 256 rest on the stratified sample plus 001's own top-8 + cross-route check. The sample covers every class and every family. +- Both float evaluators share the boundary-point construction *concept* + (support points from Hermitian eigenvectors); an error in that shared + concept would evade both. The exact-rational certificates bound this + risk: they bypass boundary sampling entirely. +- The Overton/GO paper details were read through machine renderings and + text extraction ([T]); the specific numbers re-checked here (0.698, + 0.844, run counts, methods) are as reliable as those renderings. +- 001's Lead 1 (attack the published intermediate maxima directly) is + untouched by this review; nothing here tests those basins either. + +## References + +- `problems/crouzeix/attempts/001-blind-dim3-local-maxima-census.md` (the + record under review) and its data in `problems/crouzeix/data/`. +- This review's tools and checkpoints: `explore/skeptic_recert.py`, + `explore/skeptic_probe.py`, `data/skeptic_recert.jsonl`, + `data/skeptic_probe.jsonl`, `data/skeptic_recert.log`, + `data/skeptic_probe.log`. +- M. L. Overton, *Local minimizers of the Crouzeix ratio: a nonsmooth + optimization case study*, Calcolo 58 (2021); arXiv:2105.14176 — read via + the ar5iv HTML rendering [T]; abstract page fetched directly. +- A. Greenbaum, M. L. Overton, *Numerical investigation of Crouzeix's + conjecture*, Linear Algebra Appl. 542 (2018); preprint + optimization-online 5703 (2016), PDF fetched and text-extracted for the + methods check. +- Tier-0 harness: `harness/crouzeix/ratio.py`, + `harness/crouzeix/verify_ratio.py` (audited, not reused, by this review's + certifier). diff --git a/problems/crouzeix/data/skeptic_probe.jsonl b/problems/crouzeix/data/skeptic_probe.jsonl new file mode 100644 index 0000000..f97fa76 --- /dev/null +++ b/problems/crouzeix/data/skeptic_probe.jsonl @@ -0,0 +1,9 @@ +{"kind": "stalls", "seed": 300010, "family": "random", "class": "below2-max", "ratio512": 1.9676430185679914, "best512": 1.9716588990173345, "ascent": 0.00401588044934309, "indep_base": 1.967643018569576, "indep_best": 1.9716588990173327, "verdict": "ASCENT"} +{"kind": "stalls", "seed": 604002, "family": "companion", "class": "below2-max", "ratio512": 1.96025972749273, "best512": 1.9612774409733806, "ascent": 0.001017713480650606, "indep_base": 1.9602597274927303, "indep_best": 1.9612774409724396, "verdict": "ASCENT"} +{"kind": "stalls", "seed": 406004, "family": "real", "class": "below2-max", "ratio512": 1.9588423767255672, "best512": 1.9688672519890158, "ascent": 0.01002487526344864, "indep_base": 1.9588423767383987, "indep_best": 1.9688672519892871, "verdict": "ASCENT"} +{"kind": "stalls", "seed": 504010, "family": "companion", "class": "below2-max", "ratio512": 1.7811672748407623, "best512": 1.8030426514951237, "ascent": 0.021875376654361345, "indep_base": 1.7811672748407603, "indep_best": 1.8030426514951239, "verdict": "ASCENT"} +{"kind": "maxima", "seed": 403004, "family": "nearnormal", "class": "one-exact", "ratio512": 0.9999999999820248, "best512": 0.999999999999815, "ascent": 1.779021374659351e-11, "indep_base": null, "indep_best": null, "verdict": "NO-ASCENT"} +{"kind": "maxima", "seed": 603009, "family": "nearnormal", "class": "one-exact", "ratio512": 0.9999999999813017, "best512": 0.9999999999997451, "ascent": 1.8443357951980488e-11, "indep_base": null, "indep_best": null, "verdict": "NO-ASCENT"} +{"kind": "stallcheck", "seed": 400000, "family": "random", "class": "stall", "ratio512": 1.474585189086896, "best512": 1.5271989866059983, "ascent": 0.052613797519102334, "indep_base": 1.474585189086896, "indep_best": 1.5271989866004285, "verdict": "ASCENT"} +{"kind": "stallcheck", "seed": 403000, "family": "nearnormal", "class": "stall", "ratio512": 1.6399400334716803, "best512": 1.8736813341139658, "ascent": 0.2337413006422855, "indep_base": 1.6399400334729453, "indep_best": 1.8736813340764753, "verdict": "ASCENT"} +{"kind": "stallcheck", "seed": 405000, "family": "peaks", "class": "stall", "ratio512": 1.9590367342496156, "best512": 1.9649288209773224, "ascent": 0.005892086727706802, "indep_base": 1.9590367342512611, "indep_best": 1.9649288209825182, "verdict": "ASCENT"} diff --git a/problems/crouzeix/data/skeptic_probe.log b/problems/crouzeix/data/skeptic_probe.log new file mode 100644 index 0000000..5b12801 --- /dev/null +++ b/problems/crouzeix/data/skeptic_probe.log @@ -0,0 +1,9 @@ +stalls seed=300010 below2-max v512=1.967643 best512=1.971659 ascent=+0.004016 ASCENT indep 1.967643->1.971659 +stalls seed=604002 below2-max v512=1.960260 best512=1.961277 ascent=+0.001018 ASCENT indep 1.960260->1.961277 +stalls seed=406004 below2-max v512=1.958842 best512=1.968867 ascent=+0.010025 ASCENT indep 1.958842->1.968867 +stalls seed=504010 below2-max v512=1.781167 best512=1.803043 ascent=+0.021875 ASCENT indep 1.781167->1.803043 +maxima seed=403004 one-exact v512=1.000000 best512=1.000000 ascent=+0.000000 NO-ASCENT +maxima seed=603009 one-exact v512=1.000000 best512=1.000000 ascent=+0.000000 NO-ASCENT +stallcheck seed=400000 stall v512=1.474585 best512=1.527199 ascent=+0.052614 ASCENT indep 1.474585->1.527199 +stallcheck seed=403000 stall v512=1.639940 best512=1.873681 ascent=+0.233741 ASCENT indep 1.639940->1.873681 +stallcheck seed=405000 stall v512=1.959037 best512=1.964929 ascent=+0.005892 ASCENT indep 1.959037->1.964929 diff --git a/problems/crouzeix/data/skeptic_recert.jsonl b/problems/crouzeix/data/skeptic_recert.jsonl new file mode 100644 index 0000000..2590830 --- /dev/null +++ b/problems/crouzeix/data/skeptic_recert.jsonl @@ -0,0 +1,30 @@ +{"seed": 406004, "family": "real", "class": "below2-max", "float_ratio": 1.9588423767255672, "their_lo": "1.910032463487", "their_hi_f": 2.1483211535907287, "my_lo": "1.941912540556", "my_hi": "1.956734763923", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.83} +{"seed": 504010, "family": "companion", "class": "below2-max", "float_ratio": 1.7811672748407623, "their_lo": "1.754874681196", "their_hi_f": 1.8483103413998854, "my_lo": "1.747901081165", "my_hi": "1.775139879810", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.77} +{"seed": 604002, "family": "companion", "class": "below2-max", "float_ratio": 1.96025972749273, "their_lo": "1.894167519159", "their_hi_f": 2.114330788036722, "my_lo": "1.941123520848", "my_hi": "1.950838958435", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.74} +{"seed": 300010, "family": "random", "class": "below2-max", "float_ratio": 1.9676430185679914, "their_lo": "1.919347816113", "their_hi_f": 2.1797946694067907, "my_lo": "1.939502826592", "my_hi": "1.951944117975", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.75} +{"seed": 403007, "family": "nearnormal", "class": "one-exact", "float_ratio": 0.9999999999855319, "their_lo": "0.999999874463", "their_hi_f": 1.0, "my_lo": "0.999999996160", "my_hi": "1.000000000000", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.13} +{"seed": 403004, "family": "nearnormal", "class": "one-exact", "float_ratio": 0.9999999999820248, "their_lo": "0.999999685121", "their_hi_f": 1.0, "my_lo": "0.999999997642", "my_hi": "1.000000000000", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 1.5} +{"seed": 603007, "family": "nearnormal", "class": "one-exact", "float_ratio": 0.9999999999797999, "their_lo": "0.999999769329", "their_hi_f": 1.0, "my_lo": "0.999999997118", "my_hi": "1.000000000000", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 1.99} +{"seed": 603004, "family": "nearnormal", "class": "one-exact", "float_ratio": 0.9999999999855295, "their_lo": "0.992963116512", "their_hi_f": 1.0, "my_lo": "0.999999975469", "my_hi": "1.000000000000", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 1.41} +{"seed": 503000, "family": "nearnormal", "class": "one-exact", "float_ratio": 0.9999999999897815, "their_lo": "0.999999970196", "their_hi_f": 1.0, "my_lo": "0.999999997556", "my_hi": "1.000000000000", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.1} +{"seed": 406009, "family": "real", "class": "near2-double", "float_ratio": 1.9986423314570214, "their_lo": "1.969724783391", "their_hi_f": 2.0514332060563505, "my_lo": "1.988099638981", "my_hi": "1.995294777995", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.76} +{"seed": 602004, "family": "jordan2", "class": "near2-double", "float_ratio": 1.9966914813962127, "their_lo": "1.965527148518", "their_hi_f": 2.0479545029772486, "my_lo": "1.983826465130", "my_hi": "1.991180712626", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.72} +{"seed": 606007, "family": "real", "class": "near2-double", "float_ratio": 1.9963184216629908, "their_lo": "1.965215003814", "their_hi_f": 2.0970240098074187, "my_lo": "1.980764558139", "my_hi": "1.986886508349", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.74} +{"seed": 406007, "family": "real", "class": "near2-double", "float_ratio": 1.998177627046962, "their_lo": "1.964387468289", "their_hi_f": 2.02383068589495, "my_lo": "1.981619294400", "my_hi": "1.989106217859", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.77} +{"seed": 602007, "family": "jordan2", "class": "near2-triple", "float_ratio": 1.998750459681895, "their_lo": "1.963852765627", "their_hi_f": 2.0031389201004695, "my_lo": "1.982264026145", "my_hi": "1.988247618771", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.78} +{"seed": 601004, "family": "jordan3", "class": "near2-triple", "float_ratio": 1.995203391499658, "their_lo": "1.934241437245", "their_hi_f": 2.1184892451602924, "my_lo": "1.975334568908", "my_hi": "1.990137936866", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.9} +{"seed": 401001, "family": "jordan3", "class": "near2-other", "float_ratio": 1.989554662554931, "their_lo": "1.912468145160", "their_hi_f": 2.1704192426515205, "my_lo": "1.964653755490", "my_hi": "1.983227963726", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.74} +{"seed": 401005, "family": "jordan3", "class": "near2-double", "float_ratio": 1.9858703539332918, "their_lo": "1.926703933341", "their_hi_f": 2.079426855192487, "my_lo": "1.966963738651", "my_hi": "1.979421881600", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.82} +{"seed": 502004, "family": "jordan2", "class": "near2-double", "float_ratio": 1.972615867427874, "their_lo": "1.942768247245", "their_hi_f": 2.0516934133061313, "my_lo": "1.962418198983", "my_hi": "1.968146814195", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.82} +{"seed": 406006, "family": "real", "class": "near2-double", "float_ratio": 1.9763202426022828, "their_lo": "1.938127303565", "their_hi_f": 2.070208869364135, "my_lo": "1.960092650709", "my_hi": "1.967043395032", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.8} +{"seed": 406008, "family": "real", "class": "stall", "float_ratio": 1.7689261204340256, "their_lo": "1.731392839397", "their_hi_f": 1.8962424355439171, "my_lo": "1.754874594915", "my_hi": "1.764764779028", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.78} +{"seed": 405007, "family": "peaks", "class": "stall", "float_ratio": 1.9324690744761903, "their_lo": "1.875142190385", "their_hi_f": 1.999945403456609, "my_lo": "1.913353108557", "my_hi": "1.934212823149", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.77} +{"seed": 405005, "family": "peaks", "class": "near2-double", "float_ratio": 1.9828228939265313, "their_lo": "1.942032877317", "their_hi_f": 2.0366070082542116, "my_lo": "1.958211014735", "my_hi": "1.965378731814", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.78} +{"seed": 606006, "family": "real", "class": "stall", "float_ratio": 1.6270231550005925, "their_lo": "1.549770382092", "their_hi_f": 1.7253845399998538, "my_lo": "1.611632184425", "my_hi": "1.657515060376", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.82} +{"seed": 505007, "family": "peaks", "class": "stall", "float_ratio": 1.8600022210979095, "their_lo": "1.824629580814", "their_hi_f": 1.8971032880106902, "my_lo": "1.840151972851", "my_hi": "1.859965295456", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.77} +{"seed": 406003, "family": "real", "class": "near2-double", "float_ratio": 1.9829369879028436, "their_lo": "1.957060281064", "their_hi_f": 2.045347336105532, "my_lo": "1.972872375311", "my_hi": "1.979408966241", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.74} +{"seed": 404002, "family": "companion", "class": "near2-double", "float_ratio": 1.9756485744848153, "their_lo": "1.922058830465", "their_hi_f": 2.377811567236033, "my_lo": "1.954830853152", "my_hi": "1.965919952672", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.75} +{"seed": 603008, "family": "nearnormal", "class": "stall", "float_ratio": 1.9670049320670957, "their_lo": "1.924657315842", "their_hi_f": 2.0614394357629817, "my_lo": "1.928105264250", "my_hi": "1.937069162020", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.77} +{"seed": 200005, "family": "random", "class": "stall", "float_ratio": 1.7400715415585755, "their_lo": "1.693799732289", "their_hi_f": 1.9798196707577012, "my_lo": "1.694515418292", "my_hi": "1.717738764231", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.78} +{"seed": 403000, "family": "nearnormal", "class": "stall", "float_ratio": 1.6399400334716803, "their_lo": "1.595661049569", "their_hi_f": 1.9399626159056633, "my_lo": "1.628478271734", "my_hi": "1.640464298863", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.86} +{"seed": 604008, "family": "companion", "class": "stall", "float_ratio": 1.622248131205978, "their_lo": "1.589050963019", "their_hi_f": 1.6595919027418318, "my_lo": "1.592285173794", "my_hi": "1.792936907197", "my_refutes": false, "their_refutes": false, "checks": {"their_lo_le_my_hi": true, "my_lo_le_their_hi": true, "refutes_agree": true, "my_lo_below_2": true}, "consistent": true, "sec": 2.83} diff --git a/problems/crouzeix/data/skeptic_recert.log b/problems/crouzeix/data/skeptic_recert.log new file mode 100644 index 0000000..d3f7cc2 --- /dev/null +++ b/problems/crouzeix/data/skeptic_recert.log @@ -0,0 +1,32 @@ +re-certifying 30 endpoints (4 below2-max, 5 one-exact) + [1/30] seed=406004 below2-max theirs=[1.910032,2.1483] mine=[1.941913,1.956735] OK (2.8s) + [2/30] seed=504010 below2-max theirs=[1.754874,1.8483] mine=[1.747901,1.775140] OK (2.8s) + [3/30] seed=604002 below2-max theirs=[1.894167,2.1143] mine=[1.941124,1.950839] OK (2.7s) + [4/30] seed=300010 below2-max theirs=[1.919347,2.1798] mine=[1.939503,1.951944] OK (2.7s) + [5/30] seed=403007 one-exact theirs=[0.999999,1.0000] mine=[1.000000,1.000000] OK (2.1s) + [6/30] seed=403004 one-exact theirs=[0.999999,1.0000] mine=[1.000000,1.000000] OK (1.5s) + [7/30] seed=603007 one-exact theirs=[0.999999,1.0000] mine=[1.000000,1.000000] OK (2.0s) + [8/30] seed=603004 one-exact theirs=[0.992963,1.0000] mine=[1.000000,1.000000] OK (1.4s) + [9/30] seed=503000 one-exact theirs=[0.999999,1.0000] mine=[1.000000,1.000000] OK (2.1s) + [10/30] seed=406009 near2-double theirs=[1.969724,2.0514] mine=[1.988100,1.995295] OK (2.8s) + [11/30] seed=602004 near2-double theirs=[1.965527,2.0480] mine=[1.983826,1.991181] OK (2.7s) + [12/30] seed=606007 near2-double theirs=[1.965215,2.0970] mine=[1.980765,1.986887] OK (2.7s) + [13/30] seed=406007 near2-double theirs=[1.964387,2.0238] mine=[1.981619,1.989106] OK (2.8s) + [14/30] seed=602007 near2-triple theirs=[1.963852,2.0031] mine=[1.982264,1.988248] OK (2.8s) + [15/30] seed=601004 near2-triple theirs=[1.934241,2.1185] mine=[1.975335,1.990138] OK (2.9s) + [16/30] seed=401001 near2-other theirs=[1.912468,2.1704] mine=[1.964654,1.983228] OK (2.7s) + [17/30] seed=401005 near2-double theirs=[1.926703,2.0794] mine=[1.966964,1.979422] OK (2.8s) + [18/30] seed=502004 near2-double theirs=[1.942768,2.0517] mine=[1.962418,1.968147] OK (2.8s) + [19/30] seed=406006 near2-double theirs=[1.938127,2.0702] mine=[1.960093,1.967043] OK (2.8s) + [20/30] seed=406008 stall theirs=[1.731392,1.8962] mine=[1.754875,1.764765] OK (2.8s) + [21/30] seed=405007 stall theirs=[1.875142,1.9999] mine=[1.913353,1.934213] OK (2.8s) + [22/30] seed=405005 near2-double theirs=[1.942032,2.0366] mine=[1.958211,1.965379] OK (2.8s) + [23/30] seed=606006 stall theirs=[1.549770,1.7254] mine=[1.611632,1.657515] OK (2.8s) + [24/30] seed=505007 stall theirs=[1.824629,1.8971] mine=[1.840152,1.859965] OK (2.8s) + [25/30] seed=406003 near2-double theirs=[1.957060,2.0453] mine=[1.972872,1.979409] OK (2.7s) + [26/30] seed=404002 near2-double theirs=[1.922058,2.3778] mine=[1.954831,1.965920] OK (2.7s) + [27/30] seed=603008 stall theirs=[1.924657,2.0614] mine=[1.928105,1.937069] OK (2.8s) + [28/30] seed=200005 stall theirs=[1.693799,1.9798] mine=[1.694515,1.717739] OK (2.8s) + [29/30] seed=403000 stall theirs=[1.595661,1.9400] mine=[1.628478,1.640464] OK (2.9s) + [30/30] seed=604008 stall theirs=[1.589050,1.6596] mine=[1.592285,1.792937] OK (2.8s) +done: 30 endpoints, 0 inconsistent diff --git a/problems/crouzeix/explore/skeptic_probe.py b/problems/crouzeix/explore/skeptic_probe.py new file mode 100644 index 0000000..dee0a9a --- /dev/null +++ b/problems/crouzeix/explore/skeptic_probe.py @@ -0,0 +1,229 @@ +#!/usr/bin/env python3 +"""Independent escape probes for the skeptic review of 001. + +Two things distinguish this from 001's classify.py / deep_probe.py probes: + +1. **Consistent resolution.** 001's deep probe compared candidate escape + values computed at 128 boundary samples against a baseline computed at + 512. A coarser boundary sampling UNDER-estimates max_{dW}|p| and so + OVER-estimates the ratio, which can manufacture apparent ascent. Here + every accepted ascent is re-evaluated at 512 samples against a 512-sample + baseline before it counts. + +2. **Different scheme, different RNG.** Sparse coordinate-subset Gaussian + perturbations with log-uniform scales (ridges move few coordinates), plus + jittered Nelder-Mead restarts, seed 20260804 (001 used 999 / 12345 with + dense perturbations at fixed scales). + +Also re-evaluates each claimed ascent with a second, independently written +float ratio evaluator (Jacobi eigenvalues rather than the census's trig-cubic +closed form) to guard against an evaluator artifact. + +Usage: + skeptic_probe.py stalls SEED [SEED ...] probe the 4 deep-probed stalls + skeptic_probe.py maxima SEED [SEED ...] attack claimed ratio-1 maxima + +Float-only exploration tool; verdicts here feed the review record, claims are +certified elsewhere. Standard library only. +""" + +import cmath +import json +import math +import os +import random +import sys + +HERE = os.path.dirname(os.path.abspath(__file__)) +sys.path.insert(0, HERE) +from census import objective, unpack, nelder_mead, compass # noqa: E402 + +DATA = os.path.join(HERE, "..", "data") + + +# ---------------------------------------------------------------------- +# independent float ratio evaluator: Jacobi rotations, not the trig cubic +# ---------------------------------------------------------------------- + +def jacobi_eigs(h, sweeps=30): + """Eigenvalues of a Hermitian 3x3 (python complex rows) by cyclic complex + Jacobi rotations. Shares nothing with census.herm_eigs.""" + a = [row[:] for row in h] + n = 3 + v = None + for _ in range(sweeps): + off = sum(abs(a[i][j]) ** 2 for i in range(n) for j in range(n) if i != j) + if off < 1e-26: + break + for p in range(n): + for q in range(p + 1, n): + apq = a[p][q] + if abs(apq) < 1e-18: + continue + app = a[p][p].real + aqq = a[q][q].real + # complex Jacobi: phase then angle + phase = apq / abs(apq) + tau = (aqq - app) / (2 * abs(apq)) + t = (1 if tau >= 0 else -1) / (abs(tau) + math.sqrt(1 + tau * tau)) + c = 1 / math.sqrt(1 + t * t) + s = t * c + for k in range(n): + akp = a[k][p] + akq = a[k][q] + a[k][p] = c * akp - s * (phase.conjugate() * akq) + a[k][q] = s * (phase * akp) + c * akq + for k in range(n): + apk = a[p][k] + aqk = a[q][k] + a[p][k] = c * apk - s * (phase * aqk) + a[q][k] = s * (phase.conjugate() * apk) + c * aqk + return sorted(a[i][i].real for i in range(3)) + + +def ratio_indep(x, ntheta=512): + """Crouzeix ratio via Jacobi eigenvalues throughout.""" + a, p = unpack(x) + np_ = math.sqrt(sum(abs(c) ** 2 for c in p)) + if np_ < 1e-12: + return 0.0 + p = [c / np_ for c in p] + # numerator + m = [[0j] * 3 for _ in range(3)] + power = [[1 + 0j if i == j else 0j for j in range(3)] for i in range(3)] + for c in p: + for i in range(3): + for j in range(3): + m[i][j] += c * power[i][j] + power = [[sum(power[i][k] * a[k][j] for k in range(3)) + for j in range(3)] for i in range(3)] + g = [[sum(m[k][i].conjugate() * m[k][j] for k in range(3)) + for j in range(3)] for i in range(3)] + num = math.sqrt(max(0.0, jacobi_eigs(g)[2])) + # denominator: boundary support points via Jacobi top eigenvalue + a + # fine tangential |p| sweep on the support values themselves is not + # available here, so use the same boundary-point construction idea but + # with Jacobi: for each angle take the top eigenvector by inverse power + # iteration on (lam*I - H + eps). + den = 0.0 + for k in range(ntheta): + th = 2 * math.pi * k / ntheta + e = cmath.exp(-1j * th) + h = [[(e * a[i][j] + (e * a[j][i]).conjugate()) / 2 for j in range(3)] + for i in range(3)] + lam = jacobi_eigs(h)[2] + vec = top_vec(h, lam) + z = sum(vec[i].conjugate() * a[i][j] * vec[j] + for i in range(3) for j in range(3)) + val = 0j + for c in reversed(p): + val = val * z + c + den = max(den, abs(val)) + if den < 1e-14: + return 0.0 + return num / den + + +def top_vec(h, lam, iters=3): + """Inverse iteration on (H - lam I + eps I)^-1 via 3x3 solve.""" + eps = 1e-9 * (1 + abs(lam)) + m = [[h[i][j] - ((lam + eps) if i == j else 0) for j in range(3)] + for i in range(3)] + v = [1.0 + 0j, 0.7 + 0.2j, 0.4 - 0.5j] + for _ in range(iters): + v = solve3(m, v) + nv = math.sqrt(sum(abs(x) ** 2 for x in v)) + if nv < 1e-300: + break + v = [x / nv for x in v] + return v + + +def solve3(m, b): + a = [row[:] + [bi] for row, bi in zip(m, b)] + for c in range(3): + piv = max(range(c, 3), key=lambda r: abs(a[r][c])) + a[c], a[piv] = a[piv], a[c] + if abs(a[c][c]) < 1e-300: + a[c][c] = 1e-300 + for r in range(3): + if r != c: + f = a[r][c] / a[c][c] + a[r] = [x - f * y for x, y in zip(a[r], a[c])] + return [a[i][3] / a[i][i] for i in range(3)] + + +# ---------------------------------------------------------------------- +# the probe +# ---------------------------------------------------------------------- + +def probe(x, rng, nprobe=900, nm_restarts=3, label=""): + """Search for ascent near x; every candidate is confirmed at 512-vs-512 + resolution before it counts. Returns (v512, best512, best_x).""" + v512 = objective(x, 512) + n = len(x) + cands = [] + for _ in range(nprobe): + s = 10 ** rng.uniform(-3.5, -0.7) + k = rng.choice((1, 2, 3, 5, 8, n)) + idx = rng.sample(range(n), k) + y = list(x) + for i in idx: + y[i] += rng.gauss(0, s) + fy = objective(y, 128) + if fy > v512 - 5e-4: + cands.append((fy, y)) + for _ in range(nm_restarts): + y0 = [xi + rng.gauss(0, 0.02) for xi in x] + y, fy = nelder_mead(lambda z: objective(z, 128), y0, step=0.1, + maxit=1200) + cands.append((fy, y)) + yc, vc = compass(lambda z: objective(z, 256), x, objective(x, 256), + step=0.004, maxit=1200) + cands.append((vc, yc)) + cands.sort(key=lambda t: -t[0]) + best512, best_x = v512, x + for fy, y in cands[:25]: + f512 = objective(y, 512) + if f512 > best512: + best512, best_x = f512, y + return v512, best512, best_x + + +def run(seeds, kind): + recs = {r["seed"]: r for r in + (json.loads(l) for l in open(os.path.join(DATA, "classified.jsonl")))} + rng = random.Random(20260804) + out_path = os.path.join(DATA, "skeptic_probe.jsonl") + results = [] + with open(out_path, "a") as out: + for seed in seeds: + rec = recs[int(seed)] + x = rec["x"] + v512, best, bx = probe(x, rng, label=str(seed)) + ascent = best - v512 + # confirm any ascent with the independent evaluator + ind_base = ind_best = None + if ascent > 1e-4: + ind_base = ratio_indep(x, 512) + ind_best = ratio_indep(bx, 512) + verdict = ("ASCENT" if ascent > 1e-3 else + "MARGINAL" if ascent > 1e-4 else "NO-ASCENT") + row = {"kind": kind, "seed": rec["seed"], "family": rec["family"], + "class": rec["class"], "ratio512": v512, + "best512": best, "ascent": ascent, + "indep_base": ind_base, "indep_best": ind_best, + "verdict": verdict} + results.append(row) + out.write(json.dumps(row) + "\n") + out.flush() + print(f"{kind} seed={rec['seed']} {rec['class']:11s} " + f"v512={v512:.6f} best512={best:.6f} ascent={ascent:+.6f} " + f"{verdict}" + + (f" indep {ind_base:.6f}->{ind_best:.6f}" + if ind_base is not None else ""), flush=True) + return results + + +if __name__ == "__main__": + run(sys.argv[2:], sys.argv[1]) diff --git a/problems/crouzeix/explore/skeptic_recert.py b/problems/crouzeix/explore/skeptic_recert.py new file mode 100644 index 0000000..5b540cd --- /dev/null +++ b/problems/crouzeix/explore/skeptic_recert.py @@ -0,0 +1,690 @@ +#!/usr/bin/env python3 +"""Independent re-certification of 001's endpoint enclosures (skeptic pass). + +Written for the review of attempt 001 from the PROBLEM.md verification +contract, deliberately sharing NO algorithm with the tier-0 harness: + + quantity harness route (ratio.py) this route + -------------------- ---------------------------- --------------------------- + lambda_max upper bisection on an exact LDL bisection on Sylvester's + definiteness test criterion (leading principal + minors by cofactor expansion) + lambda_max lower same bisection exact Rayleigh quotient at a + rationalized power-iteration + vector (any vector is valid) + rational sqrt bounds bisection integer math.isqrt scaling + W(A) outer polygon 32 Gaussian-integer coprime 40 rounded-trig rational + directions, |a|,|b| <= 3 directions (denominator 64), + support values rounded UP to + denominator 10^8 (safe side) + max|p|^2 on polygon interval Horner on the edges Bernstein-coefficient bounds + of an ordered convex hull on ALL pairwise vertex + segments (no hull-ordering + code at all: every hull edge + is a pairwise segment, and + every pairwise segment lies + in the hull) + +Everything on the certifying path is exact rational arithmetic. Floats are +used only to CHOOSE Rayleigh vectors and inner points; soundness never +depends on those choices. + +Soundness directions (the contract's crux): + * outer2 >= (max_{W(A)} |p|)^2 because the polygon contains W(A) + (halfplanes from upper eigenvalue bounds, offsets rounded UP, vertex + filter has enlarging slack) and |p| is subharmonic so its max over the + hull is attained on the boundary, which the pairwise segments cover. + * n2_lo <= ||p(A)||^2 by Rayleigh. Hence ratio_lo <= true ratio. + * n2_hi >= ||p(A)||^2 by Sylvester bisection; inner2 <= (max_W |p|)^2 + because inner points x*Ax/x*x lie in W(A). Hence ratio_hi >= true ratio. + * refutation test in Q: n2_lo > 4*outer2 (same decision as the contract). + +Usage: + skeptic_recert.py --selftest + skeptic_recert.py --anchors re-certify the two calibration anchors + skeptic_recert.py --sample [--out F] stratified re-certification of 001's + endpoints (reads data/classified.jsonl + and data/certified.jsonl) + +Standard library only. Deterministic (fixed RNG seed). +""" + +import argparse +import cmath +import json +import math +import os +import random +import sys +import time +from fractions import Fraction as Q + +HERE = os.path.dirname(os.path.abspath(__file__)) +DATA = os.path.join(HERE, "..", "data") + +# ---------------------------------------------------------------------- +# exact complex rationals as (re, im) Fraction pairs +# ---------------------------------------------------------------------- + +def C(re=0, im=0): + return (Q(re), Q(im)) + + +def cadd(a, b): + return (a[0] + b[0], a[1] + b[1]) + + +def csub(a, b): + return (a[0] - b[0], a[1] - b[1]) + + +def cmul(a, b): + return (a[0] * b[0] - a[1] * b[1], a[0] * b[1] + a[1] * b[0]) + + +def cconj(a): + return (a[0], -a[1]) + + +def cnorm2(a): + return a[0] * a[0] + a[1] * a[1] + + +def cdiv(a, b): + n = cnorm2(b) + if n == 0: + raise ZeroDivisionError + return ((a[0] * b[0] + a[1] * b[1]) / n, (a[1] * b[0] - a[0] * b[1]) / n) + + +ZERO, ONE = C(0), C(1) + + +def mmulq(a, b): + n = len(a) + return [[sum_c(cmul(a[i][k], b[k][j]) for k in range(n)) for j in range(n)] + for i in range(n)] + + +def sum_c(it): + re, im = Q(0), Q(0) + for x in it: + re += x[0] + im += x[1] + return (re, im) + + +def madjq(a): + n = len(a) + return [[cconj(a[j][i]) for j in range(n)] for i in range(n)] + + +def maddq(a, b): + return [[cadd(x, y) for x, y in zip(ra, rb)] for ra, rb in zip(a, b)] + + +def mscaleq(s, a): + return [[cmul(s, x) for x in row] for row in a] + + +def eyeq(n): + return [[C(1) if i == j else C(0) for j in range(n)] for i in range(n)] + + +def herm_partq(a): + aa = madjq(a) + return [[((x[0] + y[0]) / 2, (x[1] + y[1]) / 2) + for x, y in zip(ra, rb)] for ra, rb in zip(a, aa)] + + +def poly_matq(coeffs, a): + n = len(a) + out = [[C(0)] * n for _ in range(n)] + power = eyeq(n) + for c in coeffs: + out = maddq(out, mscaleq(c, power)) + power = mmulq(power, a) + return out + + +# ---------------------------------------------------------------------- +# integer-sqrt rational bounds (no bisection) +# ---------------------------------------------------------------------- + +def sqrt_upper(f): + """Rational >= sqrt(f) for f >= 0, via math.isqrt.""" + if f < 0: + raise ValueError + a, b = f.numerator, f.denominator + s = math.isqrt(a * b) + if s * s < a * b: + s += 1 + return Q(s, b) + + +def sqrt_lower(f, scale=10 ** 12): + """Rational <= sqrt(f), floor at 1/scale resolution.""" + if f <= 0: + return Q(0) + a, b = f.numerator, f.denominator + return Q(math.isqrt(a * scale * scale // b), scale) + + +def sqrt_upper_scaled(f, scale=10 ** 12): + if f < 0: + raise ValueError + a, b = f.numerator, f.denominator + num = a * scale * scale + q = -(-num // b) # ceil + s = math.isqrt(q) + if s * s < q: + s += 1 + return Q(s, scale) + + +# ---------------------------------------------------------------------- +# eigenvalue bounds: Rayleigh from below, Sylvester bisection from above +# ---------------------------------------------------------------------- + +def det_cofactor(m): + """Determinant of a small complex-rational matrix by cofactor expansion. + Deliberately not elimination and not LDL.""" + n = len(m) + if n == 1: + return m[0][0] + if n == 2: + return csub(cmul(m[0][0], m[1][1]), cmul(m[0][1], m[1][0])) + total = C(0) + for j in range(n): + minor = [[m[i][k] for k in range(n) if k != j] for i in range(1, n)] + term = cmul(m[0][j], det_cofactor(minor)) + if j % 2: + term = (-term[0], -term[1]) + total = cadd(total, term) + return total + + +def sylvester_pd(h): + """Positive definiteness via leading principal minors, all exact.""" + n = len(h) + for k in range(1, n + 1): + sub = [row[:k] for row in h[:k]] + d = det_cofactor(sub) + # Hermitian principal minors are real; the imaginary part must vanish + if d[1] != 0: + raise ArithmeticError("non-real principal minor of a Hermitian matrix") + if d[0] <= 0: + return False + return True + + +def rowsum_bound(h): + """lambda_max <= max_i sum_j |h_ij| for Hermitian h (infinity norm).""" + best = Q(0) + for row in h: + s = Q(0) + for x in row: + s += sqrt_upper(cnorm2(x)) + best = max(best, s) + return best + + +def rayleigh(h, x): + """Exact (x* h x)/(x* x) for a rational complex vector x; real for + Hermitian h. A certified LOWER bound on lambda_max for any x != 0.""" + n = len(h) + num = C(0) + for i in range(n): + for j in range(n): + num = cadd(num, cmul(cmul(cconj(x[i]), h[i][j]), x[j])) + den = sum(cnorm2(v) for v in x) + if den == 0: + return None + return num[0] / den + + +def top_vec_float(hf, iters=200): + """Float power iteration for the top eigenvector of a Hermitian matrix + given as python-complex rows. Shifted to make lambda_max dominant.""" + n = len(hf) + shift = 1.0 + max(sum(abs(x) for x in row) for row in hf) + m = [[hf[i][j] + (shift if i == j else 0.0) for j in range(n)] + for i in range(n)] + v = [complex(1.0, 0.3 * k + 0.1) for k in range(n)] + for _ in range(iters): + w = [sum(m[i][j] * v[j] for j in range(n)) for i in range(n)] + nw = math.sqrt(sum(abs(x) ** 2 for x in w)) + if nw < 1e-300: + return [1.0] + [0.0] * (n - 1) + v = [x / nw for x in w] + return v + + +def to_float_mat(h): + return [[complex(float(x[0]), float(x[1])) for x in row] for row in h] + + +def rationalize_vec(vf, den=10 ** 6): + return [C(Q(round(x.real * den), den), Q(round(x.imag * den), den)) + for x in vf] + + +def lam_max_bounds(h, tol=Q(1, 10 ** 7)): + """Certified [lo, hi] around lambda_max(h): Rayleigh below, Sylvester + bisection above.""" + x = rationalize_vec(top_vec_float(to_float_mat(h))) + lo = rayleigh(h, x) + if lo is None: + lo = -rowsum_bound(h) + hi = rowsum_bound(h) + if hi < lo: + hi = lo + tol + n = len(h) + while hi - lo > tol: + mid = (lo + hi) / 2 + shifted = [[csub(((mid, Q(0)) if i == j else C(0)), h[i][j]) + for j in range(n)] for i in range(n)] + if sylvester_pd(shifted): + hi = mid + else: + lo = mid + return lo, hi + + +# ---------------------------------------------------------------------- +# outer polygon of W(A) from rounded-trig rational directions +# ---------------------------------------------------------------------- + +def trig_directions(count=40, den=64): + dirs = [] + seen = set() + for k in range(count): + th = 2 * math.pi * (k + 0.5) / count + u = (Q(round(math.cos(th) * den), den), Q(round(math.sin(th) * den), den)) + if u == (Q(0), Q(0)) or u in seen: + continue + seen.add(u) + dirs.append(u) + return dirs + + +def support_upper(a, u, tol=Q(1, 10 ** 7), cap_den=10 ** 8): + """Upper bound on sup Re(conj(u) z) over W(A): lambda_max(Herm(conj(u)A)), + rounded UP to denominator cap_den (weakening keeps the halfplane valid).""" + h = herm_partq(mscaleq(cconj(u), a)) + _, hi = lam_max_bounds(h, tol) + return Q(-(-(hi * cap_den).numerator // (hi * cap_den).denominator), cap_den) + + +def outer_vertices(a, dirs, tol=Q(1, 10 ** 7), slack=Q(1, 10 ** 9)): + planes = [(u, support_upper(a, u, tol)) for u in dirs] + pts = set() + m = len(planes) + for i in range(m): + u1, h1 = planes[i] + for j in range(i + 1, m): + u2, h2 = planes[j] + det = u1[0] * u2[1] - u1[1] * u2[0] + if det == 0: + continue + x = (h1 * u2[1] - h2 * u1[1]) / det + y = (u1[0] * h2 - u2[0] * h1) / det + if all(u[0] * x + u[1] * y <= h + slack for u, h in planes): + pts.add((x, y)) + return sorted(pts), planes + + +# ---------------------------------------------------------------------- +# Bernstein bound for |p|^2 on segments +# ---------------------------------------------------------------------- + +def binom(n, k): + return math.comb(n, k) + + +def seg_abs2_poly(coeffs, z0, z1): + """|p(z0 + s (z1 - z0))|^2 as real rational coefficients in s, ascending.""" + d = csub(z1, z0) + cur = [C(1)] + acc = [C(0)] * len(coeffs) + for c in coeffs: + for i, t in enumerate(cur): + acc[i] = cadd(acc[i], cmul(c, t)) + nxt = [C(0)] * (len(cur) + 1) + for i, t in enumerate(cur): + nxt[i] = cadd(nxt[i], cmul(t, z0)) + nxt[i + 1] = cadd(nxt[i + 1], cmul(t, d)) + cur = nxt + deg = len(acc) + out = [Q(0)] * (2 * deg - 1) + for i in range(deg): + for j in range(deg): + out[i + j] += acc[i][0] * acc[j][0] + acc[i][1] * acc[j][1] + return out + + +def bernstein_max_01(q, pieces=8): + """Upper bound for the rational polynomial q on [0,1]: subdivide, then on + each piece the max Bernstein coefficient bounds the polynomial (Bernstein + basis is a nonnegative partition of unity).""" + n = len(q) - 1 + if n <= 0: + return q[0] if q else Q(0) + best = None + w = Q(1, pieces) + for k in range(pieces): + c = Q(k, pieces) + # shift: a'_j = sum_{m>=j} q_m C(m,j) c^(m-j) w^j + ap = [] + for j in range(n + 1): + s = Q(0) + for m in range(j, n + 1): + s += q[m] * binom(m, j) * c ** (m - j) + ap.append(s * w ** j) + for i in range(n + 1): + b = Q(0) + for j in range(i + 1): + b += Q(binom(i, j), binom(n, j)) * ap[j] + if best is None or b > best: + best = b + return best + + +def outer_max_abs2(coeffs, verts, pieces=8): + """Upper bound for |p|^2 over the hull of verts: Bernstein bound over + every pairwise segment (superset of the hull's edges, subset of the hull).""" + if not verts: + return Q(0) + if len(verts) == 1: + z = (verts[0][0], verts[0][1]) + return eval_abs2(coeffs, z) + best = Q(0) + m = len(verts) + for i in range(m): + z0 = (verts[i][0], verts[i][1]) + for j in range(i + 1, m): + z1 = (verts[j][0], verts[j][1]) + q = seg_abs2_poly(coeffs, z0, z1) + b = bernstein_max_01(q, pieces) + if b > best: + best = b + return best + + +def eval_abs2(coeffs, z): + val = C(0) + power = C(1) + for c in coeffs: + val = cadd(val, cmul(c, power)) + power = cmul(power, z) + return cnorm2(val) + + +# ---------------------------------------------------------------------- +# inner points (for the upper end of the enclosure) +# ---------------------------------------------------------------------- + +def inner_points(a, extra_float_dirs=16): + n = len(a) + vecs = [] + for i in range(n): + e = [C(0)] * n + e[i] = C(1) + vecs.append(e) + for i in range(n): + for j in range(i + 1, n): + for s in (C(1), C(-1), C(0, 1), C(0, -1)): + e = [C(0)] * n + e[i] = C(1) + e[j] = s + vecs.append(e) + af = to_float_mat(a) + for k in range(extra_float_dirs): + th = 2 * math.pi * k / extra_float_dirs + e = cmath.exp(-1j * th) + hf = [[(e * af[i][j] + (e * af[j][i]).conjugate()) / 2 + for j in range(n)] for i in range(n)] + vecs.append(rationalize_vec(top_vec_float(hf), den=10 ** 4)) + pts = [] + for x in vecs: + den = sum(cnorm2(v) for v in x) + if den == 0: + continue + num = C(0) + for i in range(n): + for j in range(n): + num = cadd(num, cmul(cmul(cconj(x[i]), a[i][j]), x[j])) + pts.append((num[0] / den, num[1] / den)) + return pts + + +# ---------------------------------------------------------------------- +# the enclosure +# ---------------------------------------------------------------------- + +def certify(a, coeffs, ndirs=40, dir_den=64, tol=Q(1, 10 ** 7), pieces=8): + m = poly_matq(coeffs, a) + gram = mmulq(madjq(m), m) + n2_lo, n2_hi = lam_max_bounds(gram, tol) + n2_lo = max(n2_lo, Q(0)) + + verts, planes = outer_vertices(a, trig_directions(ndirs, dir_den), tol) + outer2 = outer_max_abs2(coeffs, verts, pieces) + + inner2 = Q(0) + for z in inner_points(a): + # sanity: inner points must satisfy every halfplane (they are in W(A)) + v = eval_abs2(coeffs, z) + if v > inner2: + inner2 = v + + if outer2 <= 0 or inner2 <= 0: + return {"degenerate": True, "outer2": outer2, "inner2": inner2} + return { + "degenerate": False, + "norm2": (n2_lo, n2_hi), + "outer2": outer2, + "inner2": inner2, + "ratio_lo": sqrt_lower(n2_lo / outer2), + "ratio_hi": sqrt_upper_scaled(n2_hi / inner2), + "refutes": n2_lo > 4 * outer2, + "nverts": len(verts), + } + + +# ---------------------------------------------------------------------- +# endpoint rationalization -- must match 001's convention exactly +# (den-256 entrywise rounding; p normalized to unit float norm first) +# ---------------------------------------------------------------------- + +def endpoint_to_rational(x, den=256): + a = [[C(0)] * 3 for _ in range(3)] + idx = [(0, 0), (1, 1), (2, 2), (0, 1), (0, 2), (1, 2)] + for k, (i, j) in enumerate(idx): + a[i][j] = C(Q(round(x[2 * k] * den), den), Q(round(x[2 * k + 1] * den), den)) + p = [complex(x[12 + 2 * k], x[13 + 2 * k]) for k in range(4)] + nrm = math.sqrt(sum(abs(c) ** 2 for c in p)) + coeffs = [C(Q(round(c.real / nrm * den), den), Q(round(c.imag / nrm * den), den)) + for c in p] + return a, coeffs + + +# ---------------------------------------------------------------------- +# reference cases +# ---------------------------------------------------------------------- + +def jordanq(n): + m = [[C(0)] * n for _ in range(n)] + for i in range(n - 1): + m[i][i + 1] = C(1) + return m + + +def diagq(vals): + n = len(vals) + m = [[C(0)] * n for _ in range(n)] + for i, v in enumerate(vals): + m[i][i] = v + return m + + +def selftest(): + ok = True + + def check(label, good, detail=""): + nonlocal ok + ok &= good + print(f"[{label}] {detail} {'OK' if good else 'FAIL'}") + + r = certify(jordanq(2), [C(0), C(1)]) + check("J2, p=z: encloses 2", r["ratio_lo"] <= 2 <= r["ratio_hi"], + f"[{float(r['ratio_lo']):.9f}, {float(r['ratio_hi']):.9f}]") + check("J2, p=z: no false refutation", not r["refutes"]) + check("J2: ||A||^2 encloses 1", r["norm2"][0] <= 1 <= r["norm2"][1]) + + r3 = certify(jordanq(3), [C(0), C(1)]) + rt2 = Q(14142135623, 10 ** 10) + check("J3, p=z: encloses sqrt2", r3["ratio_lo"] <= rt2 <= r3["ratio_hi"], + f"[{float(r3['ratio_lo']):.9f}, {float(r3['ratio_hi']):.9f}]") + + # normal control (attack b): for diagonal A the exact denominator is + # known: max |p| over hull of the spectrum. Here A = diag(1, i, -1), + # p = z: true max = 1. The polygon must OVER-estimate (outer2 >= 1) + # and the inner points must UNDER-estimate (inner2 <= 1), so the ratio + # enclosure brackets the true ratio 1 from the safe sides. + dn = certify(diagq([C(1), C(0, 1), C(-1)]), [C(0), C(1)]) + check("normal control: outer2 >= 1 >= inner2", + dn["outer2"] >= 1 >= dn["inner2"], + f"outer2={float(dn['outer2']):.9f} inner2={float(dn['inner2']):.9f}") + check("normal control: ratio encloses 1", + dn["ratio_lo"] <= 1 <= dn["ratio_hi"], + f"[{float(dn['ratio_lo']):.9f}, {float(dn['ratio_hi']):.9f}]") + check("normal control: no refutation", not dn["refutes"]) + + # a wrong claim must be rejectable: ratio_lo of the normal case must not + # reach 2 (the enclosure can actually discriminate) + check("normal control: lower end well below 2", dn["ratio_lo"] < Q(3, 2)) + + print("SELFTEST", "PASSED" if ok else "FAILED") + return ok + + +def anchors(): + """Re-certify 001's two calibration anchors with ~80 directions.""" + out = {} + print("[anchor 1: J3, p = z^2] 001 claims [1.980557335, 2.25], true value 2") + r = certify(jordanq(3), [C(0), C(0), C(1)], ndirs=80, dir_den=128) + print(f" independent enclosure [{float(r['ratio_lo']):.9f}, " + f"{float(r['ratio_hi']):.9f}] refutes={r['refutes']}") + ok1 = r["ratio_lo"] <= 2 <= r["ratio_hi"] and not r["refutes"] \ + and abs(r["ratio_lo"] - Q(1980557335, 10 ** 9)) < Q(2, 100) + print(f" contains 2 and lower end within 0.02 of theirs: {'OK' if ok1 else 'FAIL'}") + out["J3_z2"] = (str(r["ratio_lo"]), str(r["ratio_hi"]), ok1) + + print("[anchor 2: J2 (+) [1/4], p = z] 001 claims [1.990266652, 2.0], true value 2") + a = jordanq(3) + a[0][1] = C(1) + a[1][2] = C(0) + a[2][2] = C(Q(1, 4)) + r = certify(a, [C(0), C(1)], ndirs=80, dir_den=128) + print(f" independent enclosure [{float(r['ratio_lo']):.9f}, " + f"{float(r['ratio_hi']):.9f}] refutes={r['refutes']}") + ok2 = r["ratio_lo"] <= 2 <= r["ratio_hi"] and not r["refutes"] \ + and abs(r["ratio_lo"] - Q(1990266652, 10 ** 9)) < Q(2, 100) + print(f" contains 2 and lower end within 0.02 of theirs: {'OK' if ok2 else 'FAIL'}") + out["J2c_z"] = (str(r["ratio_lo"]), str(r["ratio_hi"]), ok2) + return ok1 and ok2 + + +def pick_sample(classified, certified): + """Stratified sample: all 4 below2-max, 5 one-exact, 10 near-2 including + the certified-lower-bound record holder, the rest random. Deterministic.""" + bycl = {} + for r in classified: + bycl.setdefault(r["class"], []).append(r) + cert_by_seed = {c["seed"]: c for c in certified} + rng = random.Random(20260804) + sample = [] + sample += bycl.get("below2-max", []) + one = sorted(bycl.get("one-exact", []), key=lambda r: r["seed"]) + sample += rng.sample(one, 5) + near2 = [r for r in classified if r["class"].startswith("near2")] + near2s = sorted(near2, key=lambda r: -float(cert_by_seed[r["seed"]]["ratio_lo"])) + picked = near2s[:5] # top certified incl. record holder + picked += [r for r in near2 if r["class"] == "near2-triple" + if r not in picked][:2] + rest2 = [r for r in near2 if r not in picked] + picked += rng.sample(rest2, 10 - len(picked)) + sample += picked + remaining = [r for r in classified if r not in sample] + sample += rng.sample(remaining, 11) + return sample, cert_by_seed + + +def run_sample(outfile): + classified = [json.loads(l) for l in + open(os.path.join(DATA, "classified.jsonl"))] + certified = [json.loads(l) for l in + open(os.path.join(DATA, "certified.jsonl"))] + sample, cert_by_seed = pick_sample(classified, certified) + print(f"re-certifying {len(sample)} endpoints " + f"({[r['class'] for r in sample].count('below2-max')} below2-max, " + f"{[r['class'] for r in sample].count('one-exact')} one-exact)") + n_bad = 0 + with open(outfile, "w") as out: + for i, rec in enumerate(sample): + theirs = cert_by_seed[rec["seed"]] + a, coeffs = endpoint_to_rational(rec["x"]) + t0 = time.time() + r = certify(a, coeffs) + dt = time.time() - t0 + their_lo = Q(theirs["ratio_lo"]) + their_hi = Q(theirs["ratio_hi_f"]).limit_denominator(10 ** 12) + checks = { + # their certified lower bound must not exceed my certified + # upper bound (both bound the same true ratio) + "their_lo_le_my_hi": their_lo <= r["ratio_hi"], + # my certified lower bound must not exceed their upper end + "my_lo_le_their_hi": r["ratio_lo"] <= their_hi + Q(1, 10 ** 6), + "refutes_agree": bool(r["refutes"]) == bool(theirs["refutes"]), + "my_lo_below_2": r["ratio_lo"] < 2, + } + bad = not all(checks.values()) + n_bad += bad + row = { + "seed": rec["seed"], "family": rec["family"], + "class": rec["class"], "float_ratio": rec["diag"]["ratio"], + "their_lo": theirs["ratio_lo"], "their_hi_f": theirs["ratio_hi_f"], + "my_lo": f"{float(r['ratio_lo']):.12f}", + "my_hi": f"{float(r['ratio_hi']):.12f}", + "my_refutes": r["refutes"], "their_refutes": theirs["refutes"], + "checks": checks, "consistent": not bad, "sec": round(dt, 2), + } + out.write(json.dumps(row) + "\n") + out.flush() + print(f" [{i+1}/{len(sample)}] seed={rec['seed']} {rec['class']:12s} " + f"theirs=[{theirs['ratio_lo'][:8]},{theirs['ratio_hi_f']:.4f}] " + f"mine=[{float(r['ratio_lo']):.6f},{float(r['ratio_hi']):.6f}] " + f"{'OK' if not bad else '*** INCONSISTENT ***'} ({dt:.1f}s)") + print(f"done: {len(sample)} endpoints, {n_bad} inconsistent") + return n_bad == 0 + + +def main(): + ap = argparse.ArgumentParser(description=__doc__.splitlines()[0]) + ap.add_argument("--selftest", action="store_true") + ap.add_argument("--anchors", action="store_true") + ap.add_argument("--sample", action="store_true") + ap.add_argument("--out", default=os.path.join(DATA, "skeptic_recert.jsonl")) + args = ap.parse_args() + if args.selftest: + sys.exit(0 if selftest() else 1) + if args.anchors: + sys.exit(0 if anchors() else 1) + if args.sample: + sys.exit(0 if run_sample(args.out) else 1) + ap.print_help() + + +if __name__ == "__main__": + main() diff --git a/problems/crouzeix/prior-art.json b/problems/crouzeix/prior-art.json index c276e31..6964d24 100644 --- a/problems/crouzeix/prior-art.json +++ b/problems/crouzeix/prior-art.json @@ -1,7 +1,7 @@ { "problem": "crouzeix", "route_status": "EVIDENCE", - "route_summary": "First attempt (blind, 2026-08-04): a certified 286-start local-maxima census at n=3, deg<=3 found no unknown extremal structure; skeptic pass pending. The published intermediate GO maxima are untested by this design and are the standing gap.", + "route_summary": "First attempt (blind, 2026-08-04): a certified 286-start local-maxima census at n=3, deg<=3 found no unknown extremal structure. Skeptic pass complete (002): the evidence stands - independent re-certification 30/30 consistent, stall verdicts and ratio-1 maxima confirmed - with corrections (near-2 count is 94 not 97; the certificates prove 0-certified-above-2, not below-2; Calcolo 2021 is Overton solo). The published intermediate GO/Overton maxima are untested by this design and remain the standing gap.", "attempts": [ { "id": "001", @@ -26,7 +26,41 @@ "range": "dimension 3, polynomial degree <= 3; 286 optimization starts (families: random 81, perturbed-J3 36, perturbed-J2+c 34, near-normal 34, companion-like 34, multi-peak 34, real 33); certified enclosures at all 286 endpoints rounded to denominator 256 (tol 1e-6, 32 half-planes), 0 refutations, max certified lower bound 1.969725", "gaps": [ "published intermediate local maxima (ratios ~1.185, ~1.433 at this slice, per Greenbaum-Overton Calcolo 2021) were never reached by this design; their basins are untested here and the census says nothing about them" - ] + ], + "superseded_by": "002" + }, + { + "id": "002", + "file": "attempts/002-skeptic-review-of-001.md", + "date": "2026-08-04", + "mode": "informed", + "status": "VERIFIED_WITH_CORRECTIONS", + "verifies": "001", + "mechanism": [ + "adversarial-verification", + "independent-reimplementation", + "exact-rational-arithmetic", + "certified-enclosure", + "optimization-landscape-census" + ], + "one_line": "Independent re-implementation (Sylvester-minor bisection + Rayleigh + Bernstein segment bounds, nothing shared with the harness) confirms 001's census evidence: 30/30 stratified re-certifications consistent, both anchors reproduced to the digit, all 4 stall verdicts confirmed under a fair 512-vs-512 protocol, both attacked ratio-1 maxima survive with zero ascent; corrections: near-2 class is 94 not 97, the index's 'all 286 certified below 2' overstates 0-refutations (182/286 upper ends exceed 2), Calcolo 2021 is Overton solo, and the deep-probe protocol compared mismatched boundary resolutions (quantified harmless).", + "range": "re-certified 30 of 286 endpoints (all 4 below2-max, 5 of 13 one-exact, 13 near-2 incl. the 1.969725 record holder, 8 stalls) plus both calibration anchors and the 2x2 Jordan ceiling; independent escape probes at the 4 killed intermediates, 2 ratio-1 maxima, 3 spot-check stalls; full count reconciliation of all 286 records", + "corrections": [ + "near-2 endpoint count is 94 (48 double + 44 other + 2 triple), not 97 as in 001's prose and index one_line", + "'all 286 endpoints certified below 2' (001's index one_line) should read '0 endpoints certified above 2, max certified lower bound 1.969725'; 182/286 enclosure upper ends are >= 2", + "arXiv:2105.14176 / Calcolo 58 (2021) is by M. L. Overton alone, not Greenbaum-Overton as cited", + "deep-probe ascents were 0.0045-0.0054, not 0.005-0.006, and were measured at 128 samples against a 512-sample baseline; consistent-resolution re-probing confirms the stall verdicts regardless", + "2 of the 13 ratio-1 endpoints have a near-flat multi-peak |p| profile, not a single peak; the dominant peak is at the boundary eigenvalue in all 13" + ], + "leak_terms": [ + "94 not 97 near-2 count", + "0-certified-above-2 correction", + "Sylvester-Bernstein recertification", + "512-vs-512 fair-protocol probe", + "design-was-assigned blind caveat" + ], + "note": "001's design brief traces to STATUS.md queue item 10 (tier-1), so its blind label covers what the worker read, not the choice of approach; read the census as blind evidence about findings, not about design convergence.", + "methodological_lesson": "Escape probes must compare candidate and baseline at the same boundary resolution: coarser sampling under-estimates the denominator and manufactures apparent ascent. 001's verdicts survived (20/175 stalls had self-inflation, all still genuinely escaped), but the protocol should compare like with like." } ] -} \ No newline at end of file +} From a8c22277ec374f337cc2752b40fd535356177e3d Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 4 Aug 2026 11:32:08 +0000 Subject: [PATCH 14/14] Ledger: close the crouzeix line - cycle complete, all ten problems attempted Crouzeix moves off 'no attempts': verified-results entry for the 286-start certified census (001/002) with the skeptic's corrected scoping (0 certified above 2; near-2 class 94), queue item 10 restated to the intermediate-maxima basin hunt (informed; the blind census is spent), TL/DR finalized, counts updated (25 verified results, ~56 tools, no unattempted problems). Two insights recorded: blind-mode data point #3 with the task-framing caveat, and the escape-probe sample-parity lesson. Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_011D7zA3zXd1mMPYsjz6RXqz --- STATUS.md | 57 ++++++++++++++++++++++++++++++++++++++++++++++++------- 1 file changed, 50 insertions(+), 7 deletions(-) diff --git a/STATUS.md b/STATUS.md index 46013c9..43e4668 100644 --- a/STATUS.md +++ b/STATUS.md @@ -29,8 +29,11 @@ n = 96). Probe-before-proof vindicated a second time: certified positives at n ≤ 32 masked the asymptotic failure. Also this cycle: the **entire 005 Laurent block is now FORMALIZED** (011/012 — closed-form composition, I1–I4, glide facts, specializations; 23 theorems, zero -corrections, statement review + independent rebuild complete), and the -first blind attempt on crouzeix ran (integration in progress). +corrections, statement review + independent rebuild complete), and +**crouzeix got its first attempt** (blind, 001, skeptic-confirmed 002): +a certified 286-start local-maxima census at n = 3, deg ≤ 3 finds no +unknown extremal structure — no problem in the portfolio is +unattempted any more. To pick something up, take an item from the attempt queue and follow `docs/CYCLE.md` — either by hand, one line at a time, or by restarting an @@ -66,9 +69,10 @@ surviving; route LIVE, ceiling 0.4315 vs record 0.38271. Standing library: Singmaster census to 2.5×10^29; Erdős–Gyárfás cubics to n=22 + girth-≥5 at n=24; lonely-runner k=8; Erdős–Straus identity-poverty; graceful n ≤ 14; mahler-4d and billiards blind -censuses. As of 2026-08-04: 24 verified results, 8 recorded dead -ends, 10 problems, ~48 reusable tools, and a machine-checked -certificate layer (24 Lean theorems) over the billiards route. +censuses. As of 2026-08-04: 25 verified results, 8 recorded dead +ends, 10 problems all with at least one attempt, ~56 reusable tools, +and a machine-checked certificate layer (24 Lean theorems) over the +billiards route. ## Problem status @@ -83,7 +87,7 @@ certificate layer (24 Lean theorems) over the billiards route. | Collatz | queued | low (long shot) | failed-approach taxonomy; cycle-bound frontier | | Triangular billiards | death law CLOSED both sides: parametric necessity all (a,b), death = γ_d exactly for 20 members (005/008); 135° stall dissolved, birth law + exact-135 certificates (006/007) | high | next: parametric sufficiency + birth-law theorem (queue 11); coverage conjecture + sampler blind spot (queue 12); Lean lane: L1 + the full Laurent block FORMALIZED (009–012); the geometry bridge stays informal by design | | Mahler in ℝ⁴ | census done blind (skeptic-confirmed) | medium | next: close k=12–20 (falsifiable: no proper mask with P<11); run the same pipeline on {0,±1}³ for the n=3 spectrum comparison | -| Crouzeix | onboarded, no attempts | medium | harness ready, certification risk retired; first attempt is the dim-3 landscape (run blind) | +| Crouzeix | dim-3 census done (blind, skeptic-confirmed) | medium | next: hunt the published intermediate-maxima basins (informed; seed at Overton's ≈1.185/≈1.433 configurations) — the census's recorded gap | ## Attempt queue (next cycles pull from the top) @@ -96,7 +100,7 @@ certificate layer (24 Lean theorems) over the billiards route. 7. [erdos-gyarfas] Cycle-spectrum realizability census from the n≤20/22/24 data (which length-sets occur?) — standalone interest. 8. [collatz] Failed-approach taxonomy page (library showcase; pure writing + citation verification). 9. [graceful-trees] Mine the symmetric-spider seed (LpH?GCAO??_@?A genre) at n = 15-16 targeted; lobster verification at larger n. -10. [crouzeix] **Run blind.** Local-maxima census, dimension 3, polynomial degree ≤ 3: do all basins terminate at known extremal structure? `EVIDENCE` scoped by dimension, degree and search design, all of which must be in the record — and state how the search design differs from Greenbaum–Overton's before running it, since reproducing their finding is not a result. Kill condition (measured, not hypothetical): one certified ratio enclosure costs about 0.4s in dim 2, 0.75s in dim 3 and 1.4s in dim 4 at tolerance 1e-9 with 32 directions, so a wide census is affordable only in the low thousands of points; if the design needs more, narrow to structured families where the norm has a closed form and record the narrowing. +10. [crouzeix] **Hunt the intermediate-maxima basins** (from 001/002's recorded gap; informed — the blind census is spent). Seed local maximization AT Overton's published intermediate configurations (ratios ≈ 1.185 and ≈ 1.433 at n = 3; re-derive the seeds from arXiv:2105.14176's descriptions, not the [L] transcriptions) and map their basins with the 001 pipeline + 002's equal-sample escape probe: are they genuine local maxima under this design's probe standard, and how do their basins sit relative to the 001 start families that never found them? Falsifiable either way, and either outcome sharpens the landscape SPECULATION ({1, 2}-only) recorded in 001. 11. [billiards-triangles] **Parametric sufficiency + the birth side** (from 005/008 and 006/007, 2026-07-31): (i) prove a parametric positive lower bound on the *generic* fan-gate margins along the universal segment (α,β) = (90/a − t, 90(a−1)/(a(b+1)) + 2t), t ∈ (0, 1/4] — each margin is a 3–5-term trig polynomial with the fan index entering linearly via the prefix maps; this is the ONLY missing piece for death(W(a,b)) = γ_d(a,b) at ALL (a,b). Mind the 3-fold degenerate death corner: the gate-(2a+2) margin (identity I4) also vanishes there — a naive 2-margin Taylor route silently misses it (005). (ii) Prove the birth law γ_birth(a,b) = 180 − 90(a+b+1)/(a(b+1)) (SPECULATION; survives out-of-sample at the sampler floor incl. a > 2b+3 members) with the same gate machinery — which gate pair binds at the birth edge — and produce exact birth brackets from below (NONE exist for any member; all float births share a one-sided floor bias). Windows-touch (birth(W(a+1,a)) = death(W(a,a))) then makes the family staircase fully algebraic. Cheap side task: measure the a = 1 column, still untouched. 12. [billiards-triangles] **The coverage conjecture, and the sampler blind spot** (from 006/007, 2026-07-31; absorbs the old pinch-gap item — its motivating gap [135.000°, 135.049°] is CLOSED, W(4,3) is certified alive inside it): 006 reduced "every obtuse angle has an alive W member" to an elementary Diophantine statement (unproven; float-checked at 157 + 25 arcs over 90.5°–165°, zero failures). Prove it, using the birth law as a labelled input where needed. Note the certificates so far are POINTWISE (007's C2): window-interval continuity on sub-arcs is float + SPECULATION law only, and per-triangle coverage of a whole arc is a different (open) question — the windows are x-slivers at the corners. Separately falsifiable (007 lead): every sampler in use accumulates only at the 90/j window edges, so an interior-pinch alive window would hide from ALL current designs — build one targeted interior-accumulation test before trusting any negative screen again. 13. [mahler-4d] Close the {0,±1}⁴ universe: k = 12–20 pairs (~30M orbits at k=12, improper fraction already 77% at k=9). Falsifiable: no proper mask with k ≥ 12 has P < 11. Needs the improper-detection shortcut or a streaming canonicalizer; see 001 lead 1. Cheap side quest, same pipeline: the {0,±1}³ census for the n=3 spectrum comparison (13 pairs, trivial) — does the non-Hanner gap grow or shrink with n? @@ -385,8 +389,47 @@ certificate layer (24 Lean theorems) over the billiards route. window (≈ 0.05 at n = 96) — so the ∀λ gap closes but the window-restricted variant is untouched (queue 1). +- **[crouzeix] Dim-3 local-maxima census: no unknown extremal structure + (blind, skeptic-confirmed)** (2026-08-04, attempts 001+002): 286-start + census at n = 3, deg p ≤ 3 from seven structured start families, with + certified rational enclosures at every endpoint — scoped precisely as + **0 endpoints certified above 2** (182 enclosure upper ends exceed 2, + so this is a no-refutation statement, not a global bound). Every + probe-surviving local maximum is known structure: 13 genuine ratio-1 + maxima in the ice-cream-cone configuration (eigenvalue on ∂W(A), + |p|-peak there; probed hard by the skeptic — no ascent) and 94 + Jordan-type endpoints climbing toward ratio 2; all four intermediate + candidates confirmed as optimizer stalls by the skeptic's independent + equal-sample probe. Skeptic re-certified 30/286 with a structurally + different certifier (Sylvester-minor bisection, Bernstein segment + bounds), reproduced both calibration anchors to the last printed + digit, and confirmed the denominator discretization errs in the sound + direction. Corrections: near-2 class 94 not 97; the 2021 paper is + Overton alone. Scope: this design only — Overton's published + intermediate maxima (≈ 1.185, ≈ 1.433) never appeared and their + basins are untested (queue 10). + ## Insights / cross-problem notes +- Blind mode, data point #3 (2026-08-04, crouzeix 001/002): with prior + art physically absent, the census blind-rediscovered both known + extremal structures (the ratio-1 ice-cream-cone configuration and the + ratio-2 Jordan limit). All three blind censuses to date (mahler, + billiards, crouzeix) have now independently recovered published + structure from the harness + problem statement alone. Caveat recorded + by 002 and inherited by future blind assignments: the task *framing* + (slice, cost-first sizing, design-difference requirement) came from + the orchestrator's queue, so the blind label certifies prior-art + absence, not task-selection independence. +- Escape probes need sample-count parity (2026-08-04, crouzeix 001/002): + 001's deep probe compared 128-sample escape values against 512-sample + baselines — on a noisy nonsmooth objective that mismatch can + manufacture ascent and turn a genuine local maximum into a "stall". + 002 re-ran every stall verdict at 512-vs-512 (all survived, so the + census's conclusions stand), but the lesson generalizes: any + stall-vs-maximum verdict must compare like against like, and a + landscape census without an escape-probe pass at matched sampling + reports optimizer artifacts as mathematics. - Unit replication + re-weighting is a portable adversary genre (2026-08-04, union-closed 014/015): a family certified positive at small n can fail after replicating its adversarial core Θ(n) times