[14429] | 1 | #region License Information
|
---|
| 2 | /* HeuristicLab
|
---|
| 3 | * Copyright (C) 2002-2016 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
|
---|
| 4 | *
|
---|
| 5 | * This file is part of HeuristicLab.
|
---|
| 6 | *
|
---|
| 7 | * HeuristicLab is free software: you can redistribute it and/or modify
|
---|
| 8 | * it under the terms of the GNU General Public License as published by
|
---|
| 9 | * the Free Software Foundation, either version 3 of the License, or
|
---|
| 10 | * (at your option) any later version.
|
---|
| 11 | *
|
---|
| 12 | * HeuristicLab is distributed in the hope that it will be useful,
|
---|
| 13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
| 14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
| 15 | * GNU General Public License for more details.
|
---|
| 16 | *
|
---|
| 17 | * You should have received a copy of the GNU General Public License
|
---|
| 18 | * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
|
---|
| 19 | */
|
---|
| 20 | #endregion
|
---|
| 21 |
|
---|
| 22 | using System;
|
---|
| 23 | using System.Collections.Generic;
|
---|
| 24 | using System.Linq;
|
---|
| 25 | using System.Threading;
|
---|
| 26 | using System.Threading.Tasks;
|
---|
| 27 | using HeuristicLab.Analysis;
|
---|
| 28 | using HeuristicLab.Common;
|
---|
| 29 | using HeuristicLab.Core;
|
---|
| 30 | using HeuristicLab.Data;
|
---|
| 31 | using HeuristicLab.Optimization;
|
---|
| 32 | using HeuristicLab.Optimization.Selection;
|
---|
| 33 | using HeuristicLab.Optimization.Algorithms.SingleObjective;
|
---|
| 34 | using HeuristicLab.Optimization.Crossover;
|
---|
| 35 | using HeuristicLab.Optimization.Manipulation;
|
---|
| 36 | using HeuristicLab.Parameters;
|
---|
| 37 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
| 38 |
|
---|
| 39 | namespace HeuristicLab.Algorithms.GeneticAlgorithm {
|
---|
| 40 | public class CanonicalGeneticAlgorithm<TProblem, TEncoding, TSolution> : HeuristicAlgorithm<EvolutionaryAlgorithmContext<TProblem, TEncoding, TSolution>, TProblem, TEncoding, TSolution>
|
---|
| 41 | where TProblem : class, ISingleObjectiveProblem<TEncoding, TSolution>, ISingleObjectiveProblemDefinition<TEncoding, TSolution>
|
---|
| 42 | where TEncoding : class, IEncoding<TSolution>
|
---|
| 43 | where TSolution : class, ISolution {
|
---|
| 44 |
|
---|
| 45 | [Storable]
|
---|
| 46 | private IValueParameter<IntValue> populationSize;
|
---|
| 47 | public int PopulationSize {
|
---|
| 48 | get { return populationSize.Value.Value; }
|
---|
| 49 | set {
|
---|
| 50 | if (value < 2) throw new ArgumentException("PopulationSize cannot be smaller than 2");
|
---|
| 51 | populationSize.Value.Value = value;
|
---|
| 52 | }
|
---|
| 53 | }
|
---|
| 54 |
|
---|
| 55 | [Storable]
|
---|
| 56 | private IConstrainedValueParameter<ISelector<EvolutionaryAlgorithmContext<TProblem, TEncoding, TSolution>>> selector;
|
---|
| 57 | public IConstrainedValueParameter<ISelector<EvolutionaryAlgorithmContext<TProblem, TEncoding, TSolution>>> SelectorParameter {
|
---|
| 58 | get { return selector; }
|
---|
| 59 | }
|
---|
| 60 | public ISelector<EvolutionaryAlgorithmContext<TProblem, TEncoding, TSolution>> Selector {
|
---|
| 61 | get { return selector.Value; }
|
---|
| 62 | set {
|
---|
| 63 | if (!selector.ValidValues.Contains(value))
|
---|
| 64 | selector.ValidValues.Add(value);
|
---|
| 65 | selector.Value = value;
|
---|
| 66 | }
|
---|
| 67 | }
|
---|
| 68 |
|
---|
| 69 | [Storable]
|
---|
| 70 | private IConstrainedValueParameter<ICrossover<EvolutionaryAlgorithmContext<TProblem, TEncoding, TSolution>>> crossover;
|
---|
| 71 | public IConstrainedValueParameter<ICrossover<EvolutionaryAlgorithmContext<TProblem, TEncoding, TSolution>>> CrossoverParameter {
|
---|
| 72 | get { return crossover; }
|
---|
| 73 | }
|
---|
| 74 | public ICrossover<EvolutionaryAlgorithmContext<TProblem, TEncoding, TSolution>> Crossover {
|
---|
| 75 | get { return crossover.Value; }
|
---|
| 76 | set {
|
---|
| 77 | if (!crossover.ValidValues.Contains(value))
|
---|
| 78 | crossover.ValidValues.Add(value);
|
---|
| 79 | crossover.Value = value;
|
---|
| 80 | }
|
---|
| 81 | }
|
---|
| 82 |
|
---|
| 83 | [Storable]
|
---|
| 84 | private IConstrainedValueParameter<IManipulator<EvolutionaryAlgorithmContext<TProblem, TEncoding, TSolution>>> mutator;
|
---|
| 85 | public IConstrainedValueParameter<IManipulator<EvolutionaryAlgorithmContext<TProblem, TEncoding, TSolution>>> MutatorParameter {
|
---|
| 86 | get { return mutator; }
|
---|
| 87 | }
|
---|
| 88 | public IManipulator<EvolutionaryAlgorithmContext<TProblem, TEncoding, TSolution>> Mutator {
|
---|
| 89 | get { return mutator.Value; }
|
---|
| 90 | set {
|
---|
| 91 | if (!mutator.ValidValues.Contains(value))
|
---|
| 92 | mutator.ValidValues.Add(value);
|
---|
| 93 | mutator.Value = value;
|
---|
| 94 | }
|
---|
| 95 | }
|
---|
| 96 |
|
---|
| 97 | [Storable]
|
---|
| 98 | private IValueParameter<PercentValue> mutationProbability;
|
---|
| 99 | public double MutationProbability {
|
---|
| 100 | get { return mutationProbability.Value.Value; }
|
---|
| 101 | set { mutationProbability.Value.Value = value; }
|
---|
| 102 | }
|
---|
| 103 |
|
---|
| 104 | [Storable]
|
---|
| 105 | private IValueParameter<IntValue> elitism;
|
---|
| 106 | public int Elitism {
|
---|
| 107 | get { return elitism.Value.Value; }
|
---|
| 108 | set {
|
---|
| 109 | if (value < 0) throw new ArgumentException("Elitism cannot be negative");
|
---|
| 110 | elitism.Value.Value = value;
|
---|
| 111 | }
|
---|
| 112 | }
|
---|
| 113 |
|
---|
| 114 | [Storable]
|
---|
| 115 | protected BestAverageWorstQualityAnalyzer qualityAnalyzer;
|
---|
| 116 |
|
---|
| 117 | [StorableConstructor]
|
---|
| 118 | protected CanonicalGeneticAlgorithm(bool deserializing) : base(deserializing) { }
|
---|
| 119 | protected CanonicalGeneticAlgorithm(CanonicalGeneticAlgorithm<TProblem, TEncoding, TSolution> original, Cloner cloner)
|
---|
| 120 | : base(original, cloner) {
|
---|
| 121 | populationSize = cloner.Clone(original.populationSize);
|
---|
| 122 | selector = cloner.Clone(original.selector);
|
---|
| 123 | crossover = cloner.Clone(original.crossover);
|
---|
| 124 | mutator = cloner.Clone(original.mutator);
|
---|
| 125 | mutationProbability = cloner.Clone(original.mutationProbability);
|
---|
| 126 | elitism = cloner.Clone(original.elitism);
|
---|
| 127 | }
|
---|
| 128 | public CanonicalGeneticAlgorithm() {
|
---|
| 129 | ProblemAnalyzer = new MultiAnalyzer();
|
---|
| 130 | AlgorithmAnalyzer = new MultiAnalyzer();
|
---|
| 131 | qualityAnalyzer = new BestAverageWorstQualityAnalyzer();
|
---|
| 132 | AlgorithmAnalyzer.Operators.Add(qualityAnalyzer, true);
|
---|
| 133 |
|
---|
| 134 | Parameters.Add(populationSize = new ValueParameter<IntValue>("PopulationSize", "The number of individuals in the population", new IntValue(100)));
|
---|
| 135 | Parameters.Add(selector = new ConstrainedValueParameter<ISelector<EvolutionaryAlgorithmContext<TProblem, TEncoding, TSolution>>>("Selector", "The selection heuristic that creates the mating pool."));
|
---|
| 136 | Parameters.Add(crossover = new ConstrainedValueParameter<ICrossover<EvolutionaryAlgorithmContext<TProblem, TEncoding, TSolution>>>("Crossover", "The crossover heuristic that takes two parents and produces a child."));
|
---|
| 137 | Parameters.Add(mutator = new ConstrainedValueParameter<IManipulator<EvolutionaryAlgorithmContext<TProblem, TEncoding, TSolution>>>("Mutator", "The mutation heuristic that slightly alters an individual."));
|
---|
| 138 | Parameters.Add(mutationProbability = new ValueParameter<PercentValue>("MutationProbability", "The probability with which mutation will be applied to an individual after mating.", new PercentValue(0.05)));
|
---|
| 139 | Parameters.Add(elitism = new ValueParameter<IntValue>("Elitism", "The number of best solutions from the old population that will live in the next generation."));
|
---|
| 140 | }
|
---|
| 141 |
|
---|
| 142 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
| 143 | return new CanonicalGeneticAlgorithm<TProblem, TEncoding, TSolution>(this, cloner);
|
---|
| 144 | }
|
---|
| 145 |
|
---|
| 146 | protected override void OnProblemChanged() {
|
---|
| 147 | base.OnProblemChanged();
|
---|
| 148 | qualityAnalyzer.QualityParameter.ActualName = Problem.Evaluator.QualityParameter.ActualName;
|
---|
| 149 | qualityAnalyzer.QualityParameter.Depth = 1;
|
---|
| 150 | qualityAnalyzer.QualityParameter.Hidden = true;
|
---|
| 151 | }
|
---|
| 152 |
|
---|
| 153 | protected override void PerformInitialize(CancellationToken token) {
|
---|
| 154 | for (var i = 0; i < PopulationSize; i++) {
|
---|
| 155 | var solutionScope = CreateEmptySolutionScope();
|
---|
| 156 | Context.Scope.SubScopes.Add(solutionScope);
|
---|
| 157 | RunOperator(Problem.Encoding.SolutionCreator, solutionScope, token);
|
---|
| 158 | }
|
---|
| 159 | EvaluatePopulation(Context.Scope.SubScopes.OfType<ISingleObjectiveSolutionScope<TSolution>>().ToList(), token);
|
---|
| 160 | }
|
---|
| 161 |
|
---|
| 162 | protected override void PerformIterate(CancellationToken token) {
|
---|
| 163 | Selector.Select(Context, 2 * (PopulationSize - Elitism), true);
|
---|
| 164 | var pool = Context.MatingPool.ToList();
|
---|
| 165 |
|
---|
| 166 | var nextGen = new List<ISingleObjectiveSolutionScope<TSolution>>(PopulationSize);
|
---|
| 167 | for (var i = 0; i < PopulationSize - Elitism; i++) {
|
---|
| 168 | Context.Parents = Tuple.Create(pool[2 * i], pool[2 * i + 1]);
|
---|
| 169 | Context.Child = CreateEmptySolutionScope();
|
---|
| 170 | Crossover.Cross(Context);
|
---|
| 171 | if (Context.Random.NextDouble() < MutationProbability)
|
---|
| 172 | Mutator.Manipulate(Context);
|
---|
| 173 | nextGen.Add(Context.Child);
|
---|
| 174 | }
|
---|
| 175 | EvaluatePopulation(nextGen, token);
|
---|
| 176 | if (Elitism > 0) {
|
---|
| 177 | nextGen.AddRange(Problem.Maximization
|
---|
| 178 | ? Context.Population.OrderByDescending(x => x.Fitness).Take(Elitism)
|
---|
| 179 | : Context.Population.OrderBy(x => x.Fitness).Take(Elitism));
|
---|
| 180 | }
|
---|
| 181 |
|
---|
| 182 | Context.Scope.SubScopes.Replace(nextGen);
|
---|
| 183 | Context.Iterations++;
|
---|
| 184 | }
|
---|
| 185 |
|
---|
| 186 | private void EvaluatePopulation(ICollection<ISingleObjectiveSolutionScope<TSolution>> solutions, CancellationToken token) {
|
---|
| 187 | Parallel.ForEach(solutions, p => Evaluate(p, token));
|
---|
| 188 | Context.EvaluatedSolutions += solutions.Count;
|
---|
| 189 | var best = Problem.Maximization ? solutions.MaxItems(x => x.Fitness).First() : solutions.MinItems(x => x.Fitness).First();
|
---|
| 190 | if (IsBetter(Problem.Maximization, best.Fitness, Context.BestQuality)) {
|
---|
| 191 | Context.BestQuality = best.Fitness;
|
---|
| 192 | Context.BestSolution = (TSolution)best.Solution.Clone();
|
---|
| 193 | }
|
---|
| 194 | }
|
---|
| 195 |
|
---|
| 196 | protected override void PerformAnalyze(CancellationToken token) {
|
---|
| 197 | base.PerformAnalyze(token);
|
---|
| 198 |
|
---|
| 199 | IResult res;
|
---|
| 200 | if (!Results.TryGetValue("Generations", out res)) {
|
---|
| 201 | res = new Result("Generations", new IntValue(Context.Iterations));
|
---|
| 202 | Results.Add(res);
|
---|
| 203 | } else ((IntValue)res.Value).Value = Context.Iterations;
|
---|
| 204 | }
|
---|
| 205 | }
|
---|
| 206 | }
|
---|