java - OptaPlanner 中的详尽搜索不适用于非常简单的示例

Question

我们正在尝试创建一个简单的示例来测试 OptaPlanner 的功能。在下文中，我们展示了我们的想法。我们示例的问题在于，当我们选择穷举搜索算法来解决问题时，OptaPlanner 会迅速终止并返回错误答案，该答案始终为零，即使零不是 ValueRangeProvider 可用的可能解决方案。此外，与使用本地搜索时相反，在求解期间未设置 PlanningVariable。

我们尝试更改 OptaPlanner 附带的示例中的算法(例如 TSP)，但效果很好。因此我们的问题是:为什么我们的代码不起作用？

MyPlanningEntity.java:

import org.optaplanner.core.api.domain.entity.PlanningEntity;
import org.optaplanner.core.api.domain.variable.PlanningVariable;

@PlanningEntity
public class MyPlanningEntity {
    @PlanningVariable(valueRangeProviderRefs = {"myListValueRangeProvider"})
    private int myPlanningVariable;

    public int getMyPlanningVariable() {
        return myPlanningVariable;
    }

    public void setMyPlanningVariable(int myPlanningVariable) {
        this.myPlanningVariable = myPlanningVariable;
    }

}

我的解决方案.java:

import org.optaplanner.core.api.domain.solution.PlanningEntityProperty;
import org.optaplanner.core.api.domain.solution.PlanningSolution;
import org.optaplanner.core.api.domain.solution.Solution;
import org.optaplanner.core.api.domain.valuerange.CountableValueRange;
import org.optaplanner.core.api.domain.valuerange.ValueRangeFactory;
import org.optaplanner.core.api.domain.valuerange.ValueRangeProvider;
import org.optaplanner.core.api.score.buildin.simple.SimpleScore;

import java.util.ArrayList;
import java.util.Collection;
import java.util.List;

@PlanningSolution
public class MySolution implements Solution<SimpleScore> {
    @PlanningEntityProperty
    private MyPlanningEntity myPlanningEntity;
    private SimpleScore score;

    public MyPlanningEntity getMyPlanningEntity() {
        return myPlanningEntity;
    }

    public void setMyPlanningEntity(MyPlanningEntity myPlanningEntity) {
        this.myPlanningEntity = myPlanningEntity;
    }

    @ValueRangeProvider(id = "myListValueRangeProvider")
    public List<Integer> getListValueRange(){
        List<Integer> list = new ArrayList<>();
        list.add(1);
        list.add(2);
        list.add(3);
        list.add(4);
        return list;
    }

    @Override
    public SimpleScore getScore() {
        return score;
    }

    @Override
    public void setScore(SimpleScore simpleScore) {
        this.score = simpleScore;
    }

    @Override
    public Collection<?> getProblemFacts() {
        return null;
    }
}

MyScoreCalculator.java:

import org.optaplanner.core.api.score.Score;
import org.optaplanner.core.api.score.buildin.simple.SimpleScore;
import org.optaplanner.core.impl.score.director.easy.EasyScoreCalculator;

public class MyScoreCalculator implements EasyScoreCalculator<MySolution>{
    @Override
    public Score calculateScore(MySolution mySolution) {
        // The higher the input, the higher the output
        int value = mySolution.getMyPlanningEntity().getMyPlanningVariable();
        return SimpleScore.valueOf(value);
    }
}

ESTest.java:

import org.optaplanner.core.api.solver.Solver;
import org.optaplanner.core.api.solver.SolverFactory;

public class ESTest {
    public static void run(){
        SolverFactory solverFactory = SolverFactory.createFromXmlResource("resources/myPlanningProblem.xml");
        Solver solver = solverFactory.buildSolver();

        MySolution mySolution = new MySolution();
        MyPlanningEntity myPlanningEntity = new MyPlanningEntity();
        mySolution.setMyPlanningEntity(myPlanningEntity);

        solver.solve(mySolution);

        MySolution bestSolution = (MySolution) solver.getBestSolution();
        System.out.println("Best solution: " + bestSolution.getMyPlanningEntity().getMyPlanningVariable());
    }

    public static void main(String args[]){
        run();
    }
}

我的计划问题.xml:

<?xml version="1.0" encoding="UTF-8"?>
<solver>
    <!-- Domain model configuration -->
    <scanAnnotatedClasses/>
    <scoreDirectorFactory>
        <scoreDefinitionType>SIMPLE</scoreDefinitionType>
        <easyScoreCalculatorClass>MyScoreCalculator</easyScoreCalculatorClass>
    </scoreDirectorFactory>

    <!-- THIS DOES NOT WORK STAND ALONE -->
    <!--<constructionHeuristic>-->
        <!--<constructionHeuristicType>FIRST_FIT</constructionHeuristicType>-->
    <!--</constructionHeuristic>-->

    <!-- THIS DOES NOT WORK STAND ALONE -->
    <exhaustiveSearch>
        <exhaustiveSearchType>BRUTE_FORCE</exhaustiveSearchType>
        <termination>
            <stepCountLimit>100</stepCountLimit>
        </termination>
    </exhaustiveSearch>

    <!-- THIS WORKS BEAUTIFULLY -->
    <!--<localSearch>-->
        <!--<localSearchType>HILL_CLIMBING</localSearchType>-->
        <!--<termination>-->
            <!--<secondsSpentLimit>10</secondsSpentLimit>-->
        <!--</termination>-->
    <!--</localSearch>-->
</solver>

我们正在使用 OptaPlanner 6.3 Final。

这是使用此配置启动 OptaPlanner 时我们得到的结果:

14:58:58.742 [main] INFO  o.o.core.impl.solver.DefaultSolver - Solving started: time spent (4), best score (0), environment mode (REPRODUCIBLE), random (JDK with seed 0).
14:58:58.745 [main] INFO  o.o.c.i.e.DefaultExhaustiveSearchPhase - Exhaustive Search phase (0) ended: step total (0), time spent (7), best score (0).
14:58:58.745 [main] INFO  o.o.core.impl.solver.DefaultSolver - Solving ended: time spent (7), best score (0), average calculate count per second (285), environment mode (REPRODUCIBLE).
Best solution: 0

Process finished with exit code 0

Answer 1

规划变量应该是 Integer，而不是 int。它应该以 null 开头。 如果它以 0 开始，OptaPlanner 假定它已经初始化并且 - 默认情况下 - 不会在 CH 或 ES 中重新初始化该变量。

Maybe we should ban primitive types for planning variables?