source: trunk/HeuristicLab.Algorithms.NSGA2/3.3/NSGA2.cs @ 15815

#region License Information
/* HeuristicLab
 * Copyright (C) 2002-2018 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
 *
 * This file is part of HeuristicLab.
 *
 * HeuristicLab is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * HeuristicLab is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
 */
#endregion

using System;
using System.Linq;
using HeuristicLab.Analysis;
using HeuristicLab.Common;
using HeuristicLab.Core;
using HeuristicLab.Data;
using HeuristicLab.Operators;
using HeuristicLab.Optimization;
using HeuristicLab.Optimization.Operators;
using HeuristicLab.Parameters;
using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
using HeuristicLab.PluginInfrastructure;
using HeuristicLab.Random;

namespace HeuristicLab.Algorithms.NSGA2 {
  /// <summary>
  /// The Nondominated Sorting Genetic Algorithm II was introduced in Deb et al. 2002. A Fast and Elitist Multiobjective Genetic Algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation, 6(2), pp. 182-197.
  /// </summary>
  [Item("NSGA-II", "The Nondominated Sorting Genetic Algorithm II was introduced in Deb et al. 2002. A Fast and Elitist Multiobjective Genetic Algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation, 6(2), pp. 182-197.")]
  [Creatable(CreatableAttribute.Categories.PopulationBasedAlgorithms, Priority = 135)]
  [StorableClass]
  public class NSGA2 : HeuristicOptimizationEngineAlgorithm, IStorableContent {
    public string Filename { get; set; }

    #region Problem Properties
    public override Type ProblemType {
      get { return typeof(IMultiObjectiveHeuristicOptimizationProblem); }
    }
    public new IMultiObjectiveHeuristicOptimizationProblem Problem {
      get { return (IMultiObjectiveHeuristicOptimizationProblem)base.Problem; }
      set { base.Problem = value; }
    }
    #endregion

    #region Parameter Properties
    private ValueParameter<IntValue> SeedParameter {
      get { return (ValueParameter<IntValue>)Parameters["Seed"]; }
    }
    private ValueParameter<BoolValue> SetSeedRandomlyParameter {
      get { return (ValueParameter<BoolValue>)Parameters["SetSeedRandomly"]; }
    }
    private ValueParameter<IntValue> PopulationSizeParameter {
      get { return (ValueParameter<IntValue>)Parameters["PopulationSize"]; }
    }
    public IConstrainedValueParameter<ISelector> SelectorParameter {
      get { return (IConstrainedValueParameter<ISelector>)Parameters["Selector"]; }
    }
    private ValueParameter<PercentValue> CrossoverProbabilityParameter {
      get { return (ValueParameter<PercentValue>)Parameters["CrossoverProbability"]; }
    }
    public IConstrainedValueParameter<ICrossover> CrossoverParameter {
      get { return (IConstrainedValueParameter<ICrossover>)Parameters["Crossover"]; }
    }
    private ValueParameter<PercentValue> MutationProbabilityParameter {
      get { return (ValueParameter<PercentValue>)Parameters["MutationProbability"]; }
    }
    public IConstrainedValueParameter<IManipulator> MutatorParameter {
      get { return (IConstrainedValueParameter<IManipulator>)Parameters["Mutator"]; }
    }
    private ValueParameter<MultiAnalyzer> AnalyzerParameter {
      get { return (ValueParameter<MultiAnalyzer>)Parameters["Analyzer"]; }
    }
    private ValueParameter<IntValue> MaximumGenerationsParameter {
      get { return (ValueParameter<IntValue>)Parameters["MaximumGenerations"]; }
    }
    private ValueParameter<IntValue> SelectedParentsParameter {
      get { return (ValueParameter<IntValue>)Parameters["SelectedParents"]; }
    }

    private IFixedValueParameter<BoolValue> DominateOnEqualQualitiesParameter {
      get { return (IFixedValueParameter<BoolValue>)Parameters["DominateOnEqualQualities"]; }
    }
    #endregion

    #region Properties
    public IntValue Seed {
      get { return SeedParameter.Value; }
      set { SeedParameter.Value = value; }
    }
    public BoolValue SetSeedRandomly {
      get { return SetSeedRandomlyParameter.Value; }
      set { SetSeedRandomlyParameter.Value = value; }
    }
    public IntValue PopulationSize {
      get { return PopulationSizeParameter.Value; }
      set { PopulationSizeParameter.Value = value; }
    }
    public ISelector Selector {
      get { return SelectorParameter.Value; }
      set { SelectorParameter.Value = value; }
    }
    public PercentValue CrossoverProbability {
      get { return CrossoverProbabilityParameter.Value; }
      set { CrossoverProbabilityParameter.Value = value; }
    }
    public ICrossover Crossover {
      get { return CrossoverParameter.Value; }
      set { CrossoverParameter.Value = value; }
    }
    public PercentValue MutationProbability {
      get { return MutationProbabilityParameter.Value; }
      set { MutationProbabilityParameter.Value = value; }
    }
    public IManipulator Mutator {
      get { return MutatorParameter.Value; }
      set { MutatorParameter.Value = value; }
    }
    public MultiAnalyzer Analyzer {
      get { return AnalyzerParameter.Value; }
      set { AnalyzerParameter.Value = value; }
    }
    public IntValue MaximumGenerations {
      get { return MaximumGenerationsParameter.Value; }
      set { MaximumGenerationsParameter.Value = value; }
    }
    public IntValue SelectedParents {
      get { return SelectedParentsParameter.Value; }
      set { SelectedParentsParameter.Value = value; }
    }
    public bool DominateOnEqualQualities {
      get { return DominateOnEqualQualitiesParameter.Value.Value; }
      set { DominateOnEqualQualitiesParameter.Value.Value = value; }
    }

    private RandomCreator RandomCreator {
      get { return (RandomCreator)OperatorGraph.InitialOperator; }
    }
    private SolutionsCreator SolutionsCreator {
      get { return (SolutionsCreator)RandomCreator.Successor; }
    }
    private RankAndCrowdingSorter RankAndCrowdingSorter {
      get { return (RankAndCrowdingSorter)((SubScopesCounter)SolutionsCreator.Successor).Successor; }
    }
    private NSGA2MainLoop MainLoop {
      get { return FindMainLoop(RankAndCrowdingSorter.Successor); }
    }
    #endregion

    [Storable]
    private RankBasedParetoFrontAnalyzer paretoFrontAnalyzer;

    [StorableConstructor]
    protected NSGA2(bool deserializing) : base(deserializing) { }
    protected NSGA2(NSGA2 original, Cloner cloner)
      : base(original, cloner) {
      paretoFrontAnalyzer = (RankBasedParetoFrontAnalyzer)cloner.Clone(original.paretoFrontAnalyzer);
      AfterDeserialization();
    }
    public NSGA2() {
      Parameters.Add(new ValueParameter<IntValue>("Seed", "The random seed used to initialize the new pseudo random number generator.", new IntValue(0)));
      Parameters.Add(new ValueParameter<BoolValue>("SetSeedRandomly", "True if the random seed should be set to a random value, otherwise false.", new BoolValue(true)));
      Parameters.Add(new ValueParameter<IntValue>("PopulationSize", "The size of the population of solutions.", new IntValue(100)));
      Parameters.Add(new ConstrainedValueParameter<ISelector>("Selector", "The operator used to select solutions for reproduction."));
      Parameters.Add(new ValueParameter<PercentValue>("CrossoverProbability", "The probability that the crossover operator is applied on two parents.", new PercentValue(0.9)));
      Parameters.Add(new ConstrainedValueParameter<ICrossover>("Crossover", "The operator used to cross solutions."));
      Parameters.Add(new ValueParameter<PercentValue>("MutationProbability", "The probability that the mutation operator is applied on a solution.", new PercentValue(0.05)));
      Parameters.Add(new ConstrainedValueParameter<IManipulator>("Mutator", "The operator used to mutate solutions."));
      Parameters.Add(new ValueParameter<MultiAnalyzer>("Analyzer", "The operator used to analyze each generation.", new MultiAnalyzer()));
      Parameters.Add(new ValueParameter<IntValue>("MaximumGenerations", "The maximum number of generations which should be processed.", new IntValue(1000)));
      Parameters.Add(new ValueParameter<IntValue>("SelectedParents", "Each two parents form a new child, typically this value should be twice the population size, but because the NSGA-II is maximally elitist it can be any multiple of 2 greater than 0.", new IntValue(200)));
      Parameters.Add(new FixedValueParameter<BoolValue>("DominateOnEqualQualities", "Flag which determines wether solutions with equal quality values should be treated as dominated.", new BoolValue(false)));

      RandomCreator randomCreator = new RandomCreator();
      SolutionsCreator solutionsCreator = new SolutionsCreator();
      SubScopesCounter subScopesCounter = new SubScopesCounter();
      RankAndCrowdingSorter rankAndCrowdingSorter = new RankAndCrowdingSorter();
      ResultsCollector resultsCollector = new ResultsCollector();
      NSGA2MainLoop mainLoop = new NSGA2MainLoop();

      OperatorGraph.InitialOperator = randomCreator;

      randomCreator.RandomParameter.ActualName = "Random";
      randomCreator.SeedParameter.ActualName = SeedParameter.Name;
      randomCreator.SeedParameter.Value = null;
      randomCreator.SetSeedRandomlyParameter.ActualName = SetSeedRandomlyParameter.Name;
      randomCreator.SetSeedRandomlyParameter.Value = null;
      randomCreator.Successor = solutionsCreator;

      solutionsCreator.NumberOfSolutionsParameter.ActualName = PopulationSizeParameter.Name;
      solutionsCreator.Successor = subScopesCounter;

      subScopesCounter.Name = "Initialize EvaluatedSolutions";
      subScopesCounter.ValueParameter.ActualName = "EvaluatedSolutions";
      subScopesCounter.Successor = rankAndCrowdingSorter;

      rankAndCrowdingSorter.DominateOnEqualQualitiesParameter.ActualName = DominateOnEqualQualitiesParameter.Name;
      rankAndCrowdingSorter.CrowdingDistanceParameter.ActualName = "CrowdingDistance";
      rankAndCrowdingSorter.RankParameter.ActualName = "Rank";
      rankAndCrowdingSorter.Successor = resultsCollector;

      resultsCollector.CollectedValues.Add(new LookupParameter<IntValue>("Evaluated Solutions", null, "EvaluatedSolutions"));
      resultsCollector.ResultsParameter.ActualName = "Results";
      resultsCollector.Successor = mainLoop;

      mainLoop.PopulationSizeParameter.ActualName = PopulationSizeParameter.Name;
      mainLoop.SelectorParameter.ActualName = SelectorParameter.Name;
      mainLoop.CrossoverParameter.ActualName = CrossoverParameter.Name;
      mainLoop.CrossoverProbabilityParameter.ActualName = CrossoverProbabilityParameter.Name;
      mainLoop.MaximumGenerationsParameter.ActualName = MaximumGenerationsParameter.Name;
      mainLoop.MutatorParameter.ActualName = MutatorParameter.Name;
      mainLoop.MutationProbabilityParameter.ActualName = MutationProbabilityParameter.Name;
      mainLoop.RandomParameter.ActualName = RandomCreator.RandomParameter.ActualName;
      mainLoop.AnalyzerParameter.ActualName = AnalyzerParameter.Name;
      mainLoop.ResultsParameter.ActualName = "Results";
      mainLoop.EvaluatedSolutionsParameter.ActualName = "EvaluatedSolutions";

      foreach (ISelector selector in ApplicationManager.Manager.GetInstances<ISelector>().Where(x => !(x is ISingleObjectiveSelector)).OrderBy(x => x.Name))
        SelectorParameter.ValidValues.Add(selector);
      ISelector tournamentSelector = SelectorParameter.ValidValues.FirstOrDefault(x => x.GetType().Name.Equals("CrowdedTournamentSelector"));
      if (tournamentSelector != null) SelectorParameter.Value = tournamentSelector;

      ParameterizeSelectors();

      paretoFrontAnalyzer = new RankBasedParetoFrontAnalyzer();
      paretoFrontAnalyzer.RankParameter.ActualName = "Rank";
      paretoFrontAnalyzer.RankParameter.Depth = 1;
      paretoFrontAnalyzer.ResultsParameter.ActualName = "Results";
      ParameterizeAnalyzers();
      UpdateAnalyzers();

      AfterDeserialization();
    }

    public override IDeepCloneable Clone(Cloner cloner) {
      return new NSGA2(this, cloner);
    }

    public override void Prepare() {
      if (Problem != null) base.Prepare();
    }

    #region Events
    protected override void OnProblemChanged() {
      ParameterizeStochasticOperator(Problem.SolutionCreator);
      ParameterizeStochasticOperator(Problem.Evaluator);
      foreach (IOperator op in Problem.Operators.OfType<IOperator>()) ParameterizeStochasticOperator(op);
      ParameterizeSolutionsCreator();
      ParameterizeRankAndCrowdingSorter();
      ParameterizeMainLoop();
      ParameterizeSelectors();
      ParameterizeAnalyzers();
      ParameterizeIterationBasedOperators();
      UpdateCrossovers();
      UpdateMutators();
      UpdateAnalyzers();
      Problem.Evaluator.QualitiesParameter.ActualNameChanged += new EventHandler(Evaluator_QualitiesParameter_ActualNameChanged);
      base.OnProblemChanged();
    }
    protected override void Problem_SolutionCreatorChanged(object sender, EventArgs e) {
      ParameterizeStochasticOperator(Problem.SolutionCreator);
      ParameterizeSolutionsCreator();
      base.Problem_SolutionCreatorChanged(sender, e);
    }
    protected override void Problem_EvaluatorChanged(object sender, EventArgs e) {
      ParameterizeStochasticOperator(Problem.Evaluator);
      ParameterizeSolutionsCreator();
      ParameterizeRankAndCrowdingSorter();
      ParameterizeMainLoop();
      ParameterizeSelectors();
      ParameterizeAnalyzers();
      Problem.Evaluator.QualitiesParameter.ActualNameChanged += new EventHandler(Evaluator_QualitiesParameter_ActualNameChanged);
      base.Problem_EvaluatorChanged(sender, e);
    }
    protected override void Problem_OperatorsChanged(object sender, EventArgs e) {
      foreach (IOperator op in Problem.Operators.OfType<IOperator>()) ParameterizeStochasticOperator(op);
      ParameterizeIterationBasedOperators();
      UpdateCrossovers();
      UpdateMutators();
      UpdateAnalyzers();
      base.Problem_OperatorsChanged(sender, e);
    }
    protected override void Problem_Reset(object sender, EventArgs e) {
      base.Problem_Reset(sender, e);
    }
    private void PopulationSizeParameter_ValueChanged(object sender, EventArgs e) {
      PopulationSize.ValueChanged += new EventHandler(PopulationSize_ValueChanged);
      ParameterizeSelectors();
    }
    private void PopulationSize_ValueChanged(object sender, EventArgs e) {
      ParameterizeSelectors();
    }
    private void Evaluator_QualitiesParameter_ActualNameChanged(object sender, EventArgs e) {
      ParameterizeRankAndCrowdingSorter();
      ParameterizeMainLoop();
      ParameterizeSelectors();
      ParameterizeAnalyzers();
    }
    private void SelectedParentsParameter_ValueChanged(object sender, EventArgs e) {
      SelectedParents.ValueChanged += new EventHandler(SelectedParents_ValueChanged);
      SelectedParents_ValueChanged(null, EventArgs.Empty);
    }
    private void SelectedParents_ValueChanged(object sender, EventArgs e) {
      if (SelectedParents.Value < 2) SelectedParents.Value = 2;
      else if (SelectedParents.Value % 2 != 0) {
        SelectedParents.Value = SelectedParents.Value + 1;
      }
    }
    #endregion

    #region Helpers
    [StorableHook(HookType.AfterDeserialization)]
    private void AfterDeserialization() {
      // BackwardsCompatibility3.3
      #region Backwards compatible code, remove with 3.4
      if (!Parameters.ContainsKey("DominateOnEqualQualities"))
        Parameters.Add(new FixedValueParameter<BoolValue>("DominateOnEqualQualities", "Flag which determines wether solutions with equal quality values should be treated as dominated.", new BoolValue(false)));
      var optionalMutatorParameter = MutatorParameter as OptionalConstrainedValueParameter<IManipulator>;
      if (optionalMutatorParameter != null) {
        Parameters.Remove(optionalMutatorParameter);
        Parameters.Add(new ConstrainedValueParameter<IManipulator>("Mutator", "The operator used to mutate solutions."));
        foreach (var m in optionalMutatorParameter.ValidValues)
          MutatorParameter.ValidValues.Add(m);
        if (optionalMutatorParameter.Value == null) MutationProbability.Value = 0; // to guarantee that the old configuration results in the same behavior
        else Mutator = optionalMutatorParameter.Value;
        optionalMutatorParameter.ValidValues.Clear(); // to avoid dangling references to the old parameter its valid values are cleared
      }
      #endregion

      PopulationSizeParameter.ValueChanged += new EventHandler(PopulationSizeParameter_ValueChanged);
      PopulationSize.ValueChanged += new EventHandler(PopulationSize_ValueChanged);
      SelectedParentsParameter.ValueChanged += new EventHandler(SelectedParentsParameter_ValueChanged);
      SelectedParents.ValueChanged += new EventHandler(SelectedParents_ValueChanged);
      if (Problem != null) {
        Problem.Evaluator.QualitiesParameter.ActualNameChanged += new EventHandler(Evaluator_QualitiesParameter_ActualNameChanged);
      }
    }
    private void ParameterizeSolutionsCreator() {
      SolutionsCreator.EvaluatorParameter.ActualName = Problem.EvaluatorParameter.Name;
      SolutionsCreator.SolutionCreatorParameter.ActualName = Problem.SolutionCreatorParameter.Name;
    }
    private void ParameterizeRankAndCrowdingSorter() {
      RankAndCrowdingSorter.MaximizationParameter.ActualName = Problem.MaximizationParameter.Name;
      RankAndCrowdingSorter.QualitiesParameter.ActualName = Problem.Evaluator.QualitiesParameter.ActualName;
    }
    private void ParameterizeMainLoop() {
      MainLoop.EvaluatorParameter.ActualName = Problem.EvaluatorParameter.Name;
      MainLoop.MaximizationParameter.ActualName = Problem.MaximizationParameter.Name;
      MainLoop.QualitiesParameter.ActualName = Problem.Evaluator.QualitiesParameter.ActualName;
    }
    private void ParameterizeStochasticOperator(IOperator op) {
      if (op is IStochasticOperator)
        ((IStochasticOperator)op).RandomParameter.ActualName = RandomCreator.RandomParameter.ActualName;
    }
    private void ParameterizeSelectors() {
      foreach (ISelector selector in SelectorParameter.ValidValues) {
        selector.CopySelected = new BoolValue(true);
        selector.NumberOfSelectedSubScopesParameter.ActualName = SelectedParentsParameter.Name;
        ParameterizeStochasticOperator(selector);
      }
      if (Problem != null) {
        foreach (IMultiObjectiveSelector selector in SelectorParameter.ValidValues.OfType<IMultiObjectiveSelector>()) {
          selector.MaximizationParameter.ActualName = Problem.MaximizationParameter.Name;
          selector.QualitiesParameter.ActualName = Problem.Evaluator.QualitiesParameter.ActualName;
        }
      }
    }
    private void ParameterizeAnalyzers() {
      if (Problem != null) {
        paretoFrontAnalyzer.QualitiesParameter.ActualName = Problem.Evaluator.QualitiesParameter.ActualName;
        paretoFrontAnalyzer.QualitiesParameter.Depth = 1;
      }
    }
    private void ParameterizeIterationBasedOperators() {
      if (Problem != null) {
        foreach (IIterationBasedOperator op in Problem.Operators.OfType<IIterationBasedOperator>()) {
          op.IterationsParameter.ActualName = "Generations";
          op.MaximumIterationsParameter.ActualName = "MaximumGenerations";
        }
      }
    }
    private void UpdateCrossovers() {
      ICrossover oldCrossover = CrossoverParameter.Value;
      ICrossover defaultCrossover = Problem.Operators.OfType<ICrossover>().FirstOrDefault();
      CrossoverParameter.ValidValues.Clear();
      foreach (ICrossover crossover in Problem.Operators.OfType<ICrossover>().OrderBy(x => x.Name))
        CrossoverParameter.ValidValues.Add(crossover);
      if (oldCrossover != null) {
        ICrossover crossover = CrossoverParameter.ValidValues.FirstOrDefault(x => x.GetType() == oldCrossover.GetType());
        if (crossover != null) CrossoverParameter.Value = crossover;
        else oldCrossover = null;
      }
      if (oldCrossover == null && defaultCrossover != null)
        CrossoverParameter.Value = defaultCrossover;
    }
    private void UpdateMutators() {
      IManipulator oldMutator = MutatorParameter.Value;
      MutatorParameter.ValidValues.Clear();
      IManipulator defaultMutator = Problem.Operators.OfType<IManipulator>().FirstOrDefault();
      foreach (IManipulator mutator in Problem.Operators.OfType<IManipulator>().OrderBy(x => x.Name))
        MutatorParameter.ValidValues.Add(mutator);
      if (oldMutator != null) {
        IManipulator mutator = MutatorParameter.ValidValues.FirstOrDefault(x => x.GetType() == oldMutator.GetType());
        if (mutator != null) MutatorParameter.Value = mutator;
        else oldMutator = null;
      }
      if (oldMutator == null && defaultMutator != null)
        MutatorParameter.Value = defaultMutator;
    }
    private void UpdateAnalyzers() {
      Analyzer.Operators.Clear();
      if (Problem != null) {
        foreach (IAnalyzer analyzer in Problem.Operators.OfType<IAnalyzer>()) {
          foreach (IScopeTreeLookupParameter param in analyzer.Parameters.OfType<IScopeTreeLookupParameter>())
            param.Depth = 1;
          Analyzer.Operators.Add(analyzer, analyzer.EnabledByDefault);
        }
      }
      Analyzer.Operators.Add(paretoFrontAnalyzer, paretoFrontAnalyzer.EnabledByDefault);
    }
    private NSGA2MainLoop FindMainLoop(IOperator start) {
      IOperator mainLoop = start;
      while (mainLoop != null && !(mainLoop is NSGA2MainLoop))
        mainLoop = ((SingleSuccessorOperator)mainLoop).Successor;
      if (mainLoop == null) return null;
      else return (NSGA2MainLoop)mainLoop;
    }
    #endregion
  }
}