source: stable/HeuristicLab.Algorithms.EvolutionStrategy/3.3/EvolutionStrategyMainLoop.cs @ 18237

#region License Information
/* HeuristicLab
 * Copyright (C) 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 HeuristicLab.Common;
using HeuristicLab.Core;
using HeuristicLab.Data;
using HeuristicLab.Operators;
using HeuristicLab.Optimization.Operators;
using HeuristicLab.Parameters;
using HEAL.Attic;
using HeuristicLab.Selection;

namespace HeuristicLab.Algorithms.EvolutionStrategy {
  /// <summary>
  /// An operator which represents the main loop of an evolution strategy (EvolutionStrategy).
  /// </summary>
  [Item("EvolutionStrategyMainLoop", "An operator which represents the main loop of an evolution strategy (EvolutionStrategy).")]
  [StorableType("1B28A359-B37A-4959-A9B1-93C2E8C7C20A")]
  public sealed class EvolutionStrategyMainLoop : AlgorithmOperator {
    #region Parameter properties
    public ValueLookupParameter<IRandom> RandomParameter {
      get { return (ValueLookupParameter<IRandom>)Parameters["Random"]; }
    }
    public ValueLookupParameter<BoolValue> MaximizationParameter {
      get { return (ValueLookupParameter<BoolValue>)Parameters["Maximization"]; }
    }
    public ScopeTreeLookupParameter<DoubleValue> QualityParameter {
      get { return (ScopeTreeLookupParameter<DoubleValue>)Parameters["Quality"]; }
    }
    public ValueLookupParameter<DoubleValue> BestKnownQualityParameter {
      get { return (ValueLookupParameter<DoubleValue>)Parameters["BestKnownQuality"]; }
    }
    public ValueLookupParameter<IntValue> PopulationSizeParameter {
      get { return (ValueLookupParameter<IntValue>)Parameters["PopulationSize"]; }
    }
    public ValueLookupParameter<IntValue> ParentsPerChildParameter {
      get { return (ValueLookupParameter<IntValue>)Parameters["ParentsPerChild"]; }
    }
    public ValueLookupParameter<IntValue> ChildrenParameter {
      get { return (ValueLookupParameter<IntValue>)Parameters["Children"]; }
    }
    public ValueLookupParameter<BoolValue> PlusSelectionParameter {
      get { return (ValueLookupParameter<BoolValue>)Parameters["PlusSelection"]; }
    }
    public IValueLookupParameter<BoolValue> ReevaluateElitesParameter {
      get { return (IValueLookupParameter<BoolValue>)Parameters["ReevaluateElites"]; }
    }
    public ValueLookupParameter<IntValue> MaximumGenerationsParameter {
      get { return (ValueLookupParameter<IntValue>)Parameters["MaximumGenerations"]; }
    }
    public ValueLookupParameter<IOperator> MutatorParameter {
      get { return (ValueLookupParameter<IOperator>)Parameters["Mutator"]; }
    }
    public ValueLookupParameter<IOperator> RecombinatorParameter {
      get { return (ValueLookupParameter<IOperator>)Parameters["Recombinator"]; }
    }
    public ValueLookupParameter<IOperator> EvaluatorParameter {
      get { return (ValueLookupParameter<IOperator>)Parameters["Evaluator"]; }
    }
    public ValueLookupParameter<VariableCollection> ResultsParameter {
      get { return (ValueLookupParameter<VariableCollection>)Parameters["Results"]; }
    }
    public ValueLookupParameter<IOperator> AnalyzerParameter {
      get { return (ValueLookupParameter<IOperator>)Parameters["Analyzer"]; }
    }
    public LookupParameter<IntValue> EvaluatedSolutionsParameter {
      get { return (LookupParameter<IntValue>)Parameters["EvaluatedSolutions"]; }
    }
    private ScopeParameter CurrentScopeParameter {
      get { return (ScopeParameter)Parameters["CurrentScope"]; }
    }
    private ValueLookupParameter<IOperator> StrategyParameterManipulatorParameter {
      get { return (ValueLookupParameter<IOperator>)Parameters["StrategyParameterManipulator"]; }
    }
    private ValueLookupParameter<IOperator> StrategyParameterCrossoverParameter {
      get { return (ValueLookupParameter<IOperator>)Parameters["StrategyParameterCrossover"]; }
    }

    public IScope CurrentScope {
      get { return CurrentScopeParameter.ActualValue; }
    }
    #endregion

    [StorableConstructor]
    private EvolutionStrategyMainLoop(StorableConstructorFlag _) : base(_) { }
    private EvolutionStrategyMainLoop(EvolutionStrategyMainLoop original, Cloner cloner)
      : base(original, cloner) {
    }
    public override IDeepCloneable Clone(Cloner cloner) {
      return new EvolutionStrategyMainLoop(this, cloner);
    }
    public EvolutionStrategyMainLoop()
      : base() {
      Initialize();
    }

    [StorableHook(HookType.AfterDeserialization)]
    private void AfterDeserialization() {
      // BackwardsCompatibility3.3
      #region Backwards compatible code, remove with 3.4
      if (!Parameters.ContainsKey("ReevaluateElites")) {
        Parameters.Add(new ValueLookupParameter<BoolValue>("ReevaluateElites", "Flag to determine if elite individuals should be reevaluated (i.e., if stochastic fitness functions are used.)"));
      }
      #endregion
    }

    private void Initialize() {
      #region Create parameters
      Parameters.Add(new ValueLookupParameter<IRandom>("Random", "A pseudo random number generator."));
      Parameters.Add(new ValueLookupParameter<BoolValue>("Maximization", "True if the problem is a maximization problem, otherwise false."));
      Parameters.Add(new ScopeTreeLookupParameter<DoubleValue>("Quality", "The value which represents the quality of a solution."));
      Parameters.Add(new ValueLookupParameter<DoubleValue>("BestKnownQuality", "The best known quality value found so far."));
      Parameters.Add(new ValueLookupParameter<IntValue>("PopulationSize", "µ (mu) - the size of the population."));
      Parameters.Add(new ValueLookupParameter<IntValue>("ParentsPerChild", "ρ (rho) - how many parents should be recombined."));
      Parameters.Add(new ValueLookupParameter<IntValue>("Children", "λ (lambda) - the size of the offspring population."));
      Parameters.Add(new ValueLookupParameter<IntValue>("MaximumGenerations", "The maximum number of generations which should be processed."));
      Parameters.Add(new ValueLookupParameter<BoolValue>("PlusSelection", "True for plus selection (elitist population), false for comma selection (non-elitist population)."));
      Parameters.Add(new ValueLookupParameter<BoolValue>("ReevaluateElites", "Flag to determine if elite individuals should be reevaluated (i.e., if stochastic fitness functions are used.)"));
      Parameters.Add(new ValueLookupParameter<IOperator>("Mutator", "The operator used to mutate solutions."));
      Parameters.Add(new ValueLookupParameter<IOperator>("Recombinator", "The operator used to cross solutions."));
      Parameters.Add(new ValueLookupParameter<IOperator>("Evaluator", "The operator used to evaluate solutions. This operator is executed in parallel, if an engine is used which supports parallelization."));
      Parameters.Add(new ValueLookupParameter<VariableCollection>("Results", "The variable collection where results should be stored."));
      Parameters.Add(new ValueLookupParameter<IOperator>("Analyzer", "The operator used to analyze each generation."));
      Parameters.Add(new LookupParameter<IntValue>("EvaluatedSolutions", "The number of times solutions have been evaluated."));
      Parameters.Add(new ScopeParameter("CurrentScope", "The current scope which represents a population of solutions on which the EvolutionStrategy should be applied."));
      Parameters.Add(new ValueLookupParameter<IOperator>("StrategyParameterManipulator", "The operator to mutate the endogeneous strategy parameters."));
      Parameters.Add(new ValueLookupParameter<IOperator>("StrategyParameterCrossover", "The operator to cross the endogeneous strategy parameters."));
      #endregion

      #region Create operators
      VariableCreator variableCreator = new VariableCreator();
      ResultsCollector resultsCollector1 = new ResultsCollector();
      Placeholder analyzer1 = new Placeholder();
      WithoutRepeatingBatchedRandomSelector selector = new WithoutRepeatingBatchedRandomSelector();
      SubScopesProcessor subScopesProcessor1 = new SubScopesProcessor();
      Comparator useRecombinationComparator = new Comparator();
      ConditionalBranch useRecombinationBranch = new ConditionalBranch();
      ChildrenCreator childrenCreator = new ChildrenCreator();
      UniformSubScopesProcessor uniformSubScopesProcessor1 = new UniformSubScopesProcessor();
      Placeholder recombinator = new Placeholder();
      Placeholder strategyRecombinator = new Placeholder();
      Placeholder strategyMutator1 = new Placeholder();
      Placeholder mutator1 = new Placeholder();
      SubScopesRemover subScopesRemover = new SubScopesRemover();
      UniformSubScopesProcessor uniformSubScopesProcessor2 = new UniformSubScopesProcessor();
      Placeholder strategyMutator2 = new Placeholder();
      Placeholder mutator2 = new Placeholder();
      UniformSubScopesProcessor uniformSubScopesProcessor3 = new UniformSubScopesProcessor();
      Placeholder evaluator = new Placeholder();
      SubScopesCounter subScopesCounter = new SubScopesCounter();
      ConditionalBranch plusOrCommaReplacementBranch = new ConditionalBranch();
      MergingReducer plusReplacement = new MergingReducer();
      RightReducer commaReplacement = new RightReducer();
      BestSelector bestSelector = new BestSelector();
      RightReducer rightReducer = new RightReducer();
      IntCounter intCounter = new IntCounter();
      Comparator comparator = new Comparator();
      Placeholder analyzer2 = new Placeholder();
      ConditionalBranch conditionalBranch = new ConditionalBranch();
      ConditionalBranch reevaluateElitesBranch = new ConditionalBranch();
      SubScopesProcessor subScopesProcessor2 = new SubScopesProcessor();
      UniformSubScopesProcessor uniformSubScopesProcessor4 = new UniformSubScopesProcessor();
      Placeholder evaluator2 = new Placeholder();
      SubScopesCounter subScopesCounter2 = new SubScopesCounter();


      variableCreator.CollectedValues.Add(new ValueParameter<IntValue>("Generations", new IntValue(0))); // Class EvolutionStrategy expects this to be called Generations

      resultsCollector1.CollectedValues.Add(new LookupParameter<IntValue>("Generations"));
      resultsCollector1.ResultsParameter.ActualName = "Results";

      analyzer1.Name = "Analyzer (placeholder)";
      analyzer1.OperatorParameter.ActualName = AnalyzerParameter.Name;

      selector.Name = "ES Random Selector";
      selector.RandomParameter.ActualName = RandomParameter.Name;
      selector.ParentsPerChildParameter.ActualName = ParentsPerChildParameter.Name;
      selector.ChildrenParameter.ActualName = ChildrenParameter.Name;

      useRecombinationComparator.Name = "ParentsPerChild > 1";
      useRecombinationComparator.LeftSideParameter.ActualName = ParentsPerChildParameter.Name;
      useRecombinationComparator.RightSideParameter.Value = new IntValue(1);
      useRecombinationComparator.Comparison = new Comparison(ComparisonType.Greater);
      useRecombinationComparator.ResultParameter.ActualName = "UseRecombination";

      useRecombinationBranch.Name = "Use Recombination?";
      useRecombinationBranch.ConditionParameter.ActualName = "UseRecombination";

      childrenCreator.ParentsPerChild = null;
      childrenCreator.ParentsPerChildParameter.ActualName = ParentsPerChildParameter.Name;

      recombinator.Name = "Recombinator (placeholder)";
      recombinator.OperatorParameter.ActualName = RecombinatorParameter.Name;

      strategyRecombinator.Name = "Strategy Parameter Recombinator (placeholder)";
      strategyRecombinator.OperatorParameter.ActualName = StrategyParameterCrossoverParameter.Name;

      strategyMutator1.Name = "Strategy Parameter Manipulator (placeholder)";
      strategyMutator1.OperatorParameter.ActualName = StrategyParameterManipulatorParameter.Name;

      mutator1.Name = "Mutator (placeholder)";
      mutator1.OperatorParameter.ActualName = MutatorParameter.Name;

      subScopesRemover.RemoveAllSubScopes = true;

      strategyMutator2.Name = "Strategy Parameter Manipulator (placeholder)";
      strategyMutator2.OperatorParameter.ActualName = StrategyParameterManipulatorParameter.Name;

      mutator2.Name = "Mutator (placeholder)";
      mutator2.OperatorParameter.ActualName = MutatorParameter.Name;

      uniformSubScopesProcessor3.Parallel.Value = true;

      evaluator.Name = "Evaluator (placeholder)";
      evaluator.OperatorParameter.ActualName = EvaluatorParameter.Name;

      subScopesCounter.Name = "Increment EvaluatedSolutions";
      subScopesCounter.ValueParameter.ActualName = EvaluatedSolutionsParameter.Name;

      plusOrCommaReplacementBranch.ConditionParameter.ActualName = PlusSelectionParameter.Name;

      bestSelector.CopySelected = new BoolValue(false);
      bestSelector.MaximizationParameter.ActualName = MaximizationParameter.Name;
      bestSelector.NumberOfSelectedSubScopesParameter.ActualName = PopulationSizeParameter.Name;
      bestSelector.QualityParameter.ActualName = QualityParameter.Name;

      intCounter.Increment = new IntValue(1);
      intCounter.ValueParameter.ActualName = "Generations";

      comparator.Comparison = new Comparison(ComparisonType.GreaterOrEqual);
      comparator.LeftSideParameter.ActualName = "Generations";
      comparator.ResultParameter.ActualName = "Terminate";
      comparator.RightSideParameter.ActualName = MaximumGenerationsParameter.Name;

      analyzer2.Name = "Analyzer (placeholder)";
      analyzer2.OperatorParameter.ActualName = AnalyzerParameter.Name;

      conditionalBranch.ConditionParameter.ActualName = "Terminate";

      reevaluateElitesBranch.ConditionParameter.ActualName = "ReevaluateElites";
      reevaluateElitesBranch.Name = "Reevaluate elites ?";

      uniformSubScopesProcessor4.Parallel.Value = true;

      evaluator2.Name = "Evaluator (placeholder)";
      evaluator2.OperatorParameter.ActualName = EvaluatorParameter.Name;

      subScopesCounter2.Name = "Increment EvaluatedSolutions";
      subScopesCounter2.ValueParameter.ActualName = EvaluatedSolutionsParameter.Name;
      #endregion

      #region Create operator graph
      OperatorGraph.InitialOperator = variableCreator;
      variableCreator.Successor = resultsCollector1;
      resultsCollector1.Successor = analyzer1;
      analyzer1.Successor = selector;
      selector.Successor = subScopesProcessor1;
      subScopesProcessor1.Operators.Add(new EmptyOperator());
      subScopesProcessor1.Operators.Add(useRecombinationComparator);
      subScopesProcessor1.Successor = plusOrCommaReplacementBranch;
      useRecombinationComparator.Successor = useRecombinationBranch;
      useRecombinationBranch.TrueBranch = childrenCreator;
      useRecombinationBranch.FalseBranch = uniformSubScopesProcessor2;
      useRecombinationBranch.Successor = uniformSubScopesProcessor3;
      childrenCreator.Successor = uniformSubScopesProcessor1;
      uniformSubScopesProcessor1.Operator = recombinator;
      uniformSubScopesProcessor1.Successor = null;
      recombinator.Successor = strategyRecombinator;
      strategyRecombinator.Successor = strategyMutator1;
      strategyMutator1.Successor = mutator1;
      mutator1.Successor = subScopesRemover;
      subScopesRemover.Successor = null;
      uniformSubScopesProcessor2.Operator = strategyMutator2;
      uniformSubScopesProcessor2.Successor = null;
      strategyMutator2.Successor = mutator2;
      mutator2.Successor = null;
      uniformSubScopesProcessor3.Operator = evaluator;
      uniformSubScopesProcessor3.Successor = subScopesCounter;
      evaluator.Successor = null;
      subScopesCounter.Successor = null;

      plusOrCommaReplacementBranch.TrueBranch = reevaluateElitesBranch;
      reevaluateElitesBranch.TrueBranch = subScopesProcessor2;
      reevaluateElitesBranch.FalseBranch = null;
      subScopesProcessor2.Operators.Add(uniformSubScopesProcessor4);
      subScopesProcessor2.Operators.Add(new EmptyOperator());
      uniformSubScopesProcessor4.Operator = evaluator2;
      uniformSubScopesProcessor4.Successor = subScopesCounter2;
      subScopesCounter2.Successor = null;
      reevaluateElitesBranch.Successor = plusReplacement;

      plusOrCommaReplacementBranch.FalseBranch = commaReplacement;
      plusOrCommaReplacementBranch.Successor = bestSelector;
      bestSelector.Successor = rightReducer;
      rightReducer.Successor = intCounter;
      intCounter.Successor = comparator;
      comparator.Successor = analyzer2;
      analyzer2.Successor = conditionalBranch;
      conditionalBranch.FalseBranch = selector;
      conditionalBranch.TrueBranch = null;
      conditionalBranch.Successor = null;
      #endregion
    }

    public override IOperation Apply() {
      if (MutatorParameter.ActualValue == null)
        return null;
      return base.Apply();
    }
  }
}