solverConfig.xml

<?xml version="1.0" encoding="UTF-8"?>

<solver>
    <solutionClass>org.acme.schooltimetabling.domain.Timetable</solutionClass>
    <entityClass>org.acme.schooltimetabling.domain.lesson.Lesson</entityClass>
    <scoreDirectorFactory>
        <constraintProviderClass>org.acme.schooltimetabling.solver.TimetableConstraintProvider</constraintProviderClass>
    </scoreDirectorFactory>
    <!--Random seed is used for heuristics and meta heuristics; setting it doesn't really matter I think-->
    <randomSeed>auto</randomSeed>
    <!--stricter modes cause the solver to become slower, but can be used in dev for error-detection;
    use NON_REPRODUCIBLE if you don't reproducibility. For dev use PHASE_ASSERT (the default) or just comment
     out the environment mode; see https://docs.timefold.ai/timefold-solver/latest/using-timefold-solver/running-the-solver#environmentModeBestPractices-->
<!--    <environmentMode>NON_REPRODUCIBLE</environmentMode>-->
    <termination>
        <secondsSpentLimit>100</secondsSpentLimit>
    </termination>

    <constructionHeuristic>
        <constructionHeuristicType>FIRST_FIT_DECREASING</constructionHeuristicType>
        <entitySorterManner>DESCENDING_IF_AVAILABLE</entitySorterManner>

<!--    Cartesian product scales poorly, but I doubt it will ever be a problem for something of our size,
        However, if it ever does, just remove it and you still get a good starting state-->
        <cartesianProductMoveSelector>
            <changeMoveSelector>
                <valueSelector variableName="timeslot"/>
            </changeMoveSelector>
            <changeMoveSelector>
                <valueSelector variableName="room"/>
            </changeMoveSelector>
        </cartesianProductMoveSelector>
    </constructionHeuristic>

    <!--This phase is meant to get to a solution that has no hard score violations-->
    <localSearch>
<!--    terminate after all hard constraints are solved, with any medium or soft score-->
        <termination>
            <terminationCompositionStyle>OR</terminationCompositionStyle>
            <bestScoreLimit>0hard/*medium/*soft</bestScoreLimit>
            <unimprovedSecondsSpentLimit>30</unimprovedSecondsSpentLimit>
        </termination>
        <acceptor>
<!--        adding entity tabu helps a lot, the lower, the better. Using entity size rather than ratio is better
            from my experiments. Upon the semester switch adding tabu could harm the solution quality-->
            <entityTabuSize>1</entityTabuSize>
            <simulatedAnnealingStartingTemperature>2hard/10medium/20soft</simulatedAnnealingStartingTemperature>
        </acceptor>

<!--    from experiments, the lower this number the better, but too low seems to end up in hard constraints violated-->
<!--    10 seems to be around the sweet spot   -->
        <forager>
            <acceptedCountLimit>10</acceptedCountLimit>
        </forager>
    </localSearch>

    <!--This phase is for improving the medium score. In other words, trying to get professors their time
     preferences-->
    <localSearch>
        <termination>
            <terminationCompositionStyle>OR</terminationCompositionStyle>
            <diminishedReturns />
            <unimprovedSecondsSpentLimit>30</unimprovedSecondsSpentLimit>
            <!--The solver already heavily favors time outside primetime, so we only look to increase
            the medium score-->
            <unimprovedScoreDifferenceThreshold>0hard/1medium/*soft</unimprovedScoreDifferenceThreshold>
        </termination>
        <acceptor>
            <!--I honestly can't tell if adding in Tabu really makes a difference; I think it's better off w/o it-->
            <entityTabuSize>1</entityTabuSize>
            <!--using a ration for the values throws an error currently-->
            <valueTabuSize>3</valueTabuSize>
            <simulatedAnnealingStartingTemperature>3hard/5medium/5soft</simulatedAnnealingStartingTemperature>
        </acceptor>
        <forager>
            <acceptedCountLimit>10</acceptedCountLimit>
        </forager>
    </localSearch>


</solver>