[14420] | 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;
|
---|
[14450] | 26 | using HeuristicLab.Algorithms.MemPR.Interfaces;
|
---|
[14420] | 27 | using HeuristicLab.Common;
|
---|
| 28 | using HeuristicLab.Core;
|
---|
| 29 | using HeuristicLab.Encodings.BinaryVectorEncoding;
|
---|
| 30 | using HeuristicLab.Optimization;
|
---|
| 31 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
| 32 | using HeuristicLab.PluginInfrastructure;
|
---|
| 33 | using HeuristicLab.Random;
|
---|
| 34 |
|
---|
| 35 | namespace HeuristicLab.Algorithms.MemPR.Binary {
|
---|
| 36 | [Item("MemPR (binary)", "MemPR implementation for binary vectors.")]
|
---|
| 37 | [StorableClass]
|
---|
| 38 | [Creatable(CreatableAttribute.Categories.PopulationBasedAlgorithms, Priority = 999)]
|
---|
[14450] | 39 | public class BinaryMemPR : MemPRAlgorithm<SingleObjectiveBasicProblem<BinaryVectorEncoding>, BinaryVector, BinaryMemPRPopulationContext, BinaryMemPRSolutionContext> {
|
---|
[14420] | 40 | [StorableConstructor]
|
---|
| 41 | protected BinaryMemPR(bool deserializing) : base(deserializing) { }
|
---|
| 42 | protected BinaryMemPR(BinaryMemPR original, Cloner cloner) : base(original, cloner) { }
|
---|
| 43 | public BinaryMemPR() {
|
---|
[14450] | 44 | foreach (var trainer in ApplicationManager.Manager.GetInstances<ISolutionModelTrainer<BinaryMemPRPopulationContext>>())
|
---|
[14420] | 45 | SolutionModelTrainerParameter.ValidValues.Add(trainer);
|
---|
| 46 |
|
---|
[14450] | 47 | foreach (var localSearch in ApplicationManager.Manager.GetInstances<ILocalSearch<BinaryMemPRSolutionContext>>()) {
|
---|
| 48 | LocalSearchParameter.ValidValues.Add(localSearch);
|
---|
[14420] | 49 | }
|
---|
| 50 | }
|
---|
| 51 |
|
---|
| 52 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
| 53 | return new BinaryMemPR(this, cloner);
|
---|
| 54 | }
|
---|
| 55 |
|
---|
| 56 | protected override bool Eq(ISingleObjectiveSolutionScope<BinaryVector> a, ISingleObjectiveSolutionScope<BinaryVector> b) {
|
---|
| 57 | var len = a.Solution.Length;
|
---|
| 58 | var acode = a.Solution;
|
---|
| 59 | var bcode = b.Solution;
|
---|
| 60 | for (var i = 0; i < len; i++)
|
---|
| 61 | if (acode[i] != bcode[i]) return false;
|
---|
| 62 | return true;
|
---|
| 63 | }
|
---|
| 64 |
|
---|
| 65 | protected override double Dist(ISingleObjectiveSolutionScope<BinaryVector> a, ISingleObjectiveSolutionScope<BinaryVector> b) {
|
---|
[14496] | 66 | return 1.0 - HammingSimilarityCalculator.CalculateSimilarity(a.Solution, b.Solution);
|
---|
[14420] | 67 | }
|
---|
| 68 |
|
---|
| 69 | protected override ISingleObjectiveSolutionScope<BinaryVector> ToScope(BinaryVector code, double fitness = double.NaN) {
|
---|
| 70 | var creator = Problem.SolutionCreator as IBinaryVectorCreator;
|
---|
| 71 | if (creator == null) throw new InvalidOperationException("Can only solve binary encoded problems with MemPR (binary)");
|
---|
| 72 | return new SingleObjectiveSolutionScope<BinaryVector>(code, creator.BinaryVectorParameter.ActualName, fitness, Problem.Evaluator.QualityParameter.ActualName) {
|
---|
| 73 | Parent = Context.Scope
|
---|
| 74 | };
|
---|
| 75 | }
|
---|
| 76 |
|
---|
[14450] | 77 | protected override ISolutionSubspace<BinaryVector> CalculateSubspace(IEnumerable<BinaryVector> solutions, bool inverse = false) {
|
---|
[14420] | 78 | var pop = solutions.ToList();
|
---|
| 79 | var N = pop[0].Length;
|
---|
| 80 | var subspace = new bool[N];
|
---|
| 81 | for (var i = 0; i < N; i++) {
|
---|
| 82 | var val = pop[0][i];
|
---|
| 83 | if (inverse) subspace[i] = true;
|
---|
| 84 | for (var p = 1; p < pop.Count; p++) {
|
---|
| 85 | if (pop[p][i] != val) subspace[i] = !inverse;
|
---|
| 86 | }
|
---|
| 87 | }
|
---|
| 88 | return new BinarySolutionSubspace(subspace);
|
---|
| 89 | }
|
---|
| 90 |
|
---|
[14544] | 91 | protected override void AdaptiveWalk(ISingleObjectiveSolutionScope<BinaryVector> scope, int maxEvals, CancellationToken token, ISolutionSubspace<BinaryVector> subspace = null) {
|
---|
[14453] | 92 | var evaluations = 0;
|
---|
[14420] | 93 | var subset = subspace != null ? ((BinarySolutionSubspace)subspace).Subspace : null;
|
---|
[14453] | 94 | if (double.IsNaN(scope.Fitness)) {
|
---|
| 95 | Evaluate(scope, token);
|
---|
| 96 | evaluations++;
|
---|
| 97 | }
|
---|
[14420] | 98 | SingleObjectiveSolutionScope<BinaryVector> bestOfTheWalk = null;
|
---|
| 99 | var currentScope = (SingleObjectiveSolutionScope<BinaryVector>)scope.Clone();
|
---|
| 100 | var current = currentScope.Solution;
|
---|
| 101 | var N = current.Length;
|
---|
[14544] | 102 |
|
---|
[14420] | 103 | var subN = subset != null ? subset.Count(x => x) : N;
|
---|
[14544] | 104 | if (subN == 0) return;
|
---|
[14420] | 105 | var order = Enumerable.Range(0, N).Where(x => subset == null || subset[x]).Shuffle(Context.Random).ToArray();
|
---|
| 106 |
|
---|
[14544] | 107 | var max = Context.Population.Max(x => x.Fitness);
|
---|
| 108 | var min = Context.Population.Min(x => x.Fitness);
|
---|
| 109 | var range = (max - min);
|
---|
| 110 | if (range.IsAlmost(0)) range = Math.Abs(max * 0.05);
|
---|
| 111 | //else range += range * 0.05;
|
---|
| 112 | if (range.IsAlmost(0)) { // because min = max = 0
|
---|
| 113 | Context.IncrementEvaluatedSolutions(evaluations);
|
---|
| 114 | return;
|
---|
| 115 | }
|
---|
| 116 |
|
---|
| 117 | //var upperBound = Problem.Maximization ? min - range : max + range;
|
---|
| 118 | //var lowerBound = Problem.Maximization ? max : min;
|
---|
| 119 | var temp = 1.0;
|
---|
[14456] | 120 | for (var iter = 0; iter < int.MaxValue; iter++) {
|
---|
[14544] | 121 | var moved = false;
|
---|
[14420] | 122 |
|
---|
| 123 | for (var i = 0; i < subN; i++) {
|
---|
| 124 | var idx = order[i];
|
---|
| 125 | var before = currentScope.Fitness;
|
---|
| 126 | current[idx] = !current[idx];
|
---|
| 127 | Evaluate(currentScope, token);
|
---|
[14453] | 128 | evaluations++;
|
---|
[14420] | 129 | var after = currentScope.Fitness;
|
---|
| 130 |
|
---|
[14544] | 131 | if (Context.IsBetter(after, before) && (bestOfTheWalk == null || Context.IsBetter(after, bestOfTheWalk.Fitness))) {
|
---|
[14420] | 132 | bestOfTheWalk = (SingleObjectiveSolutionScope<BinaryVector>)currentScope.Clone();
|
---|
| 133 | }
|
---|
[14544] | 134 | var diff = Problem.Maximization ? after - before : before - after;
|
---|
| 135 | if (diff > 0) moved = true;
|
---|
| 136 | else {
|
---|
| 137 | var prob = Math.Exp(temp * diff / range);
|
---|
| 138 | if (Context.Random.NextDouble() >= prob) {
|
---|
| 139 | // the move is not good enough -> undo the move
|
---|
| 140 | current[idx] = !current[idx];
|
---|
| 141 | currentScope.Fitness = before;
|
---|
[14420] | 142 | }
|
---|
| 143 | }
|
---|
[14544] | 144 | temp += 100.0 / maxEvals;
|
---|
[14456] | 145 | if (evaluations >= maxEvals) break;
|
---|
[14420] | 146 | }
|
---|
[14544] | 147 | if (!moved) break;
|
---|
[14456] | 148 | if (evaluations >= maxEvals) break;
|
---|
[14420] | 149 | }
|
---|
| 150 |
|
---|
[14453] | 151 | Context.IncrementEvaluatedSolutions(evaluations);
|
---|
[14420] | 152 | scope.Adopt(bestOfTheWalk ?? currentScope);
|
---|
| 153 | }
|
---|
| 154 |
|
---|
[14544] | 155 | protected override ISingleObjectiveSolutionScope<BinaryVector> Breed(ISingleObjectiveSolutionScope<BinaryVector> p1, ISingleObjectiveSolutionScope<BinaryVector> p2, CancellationToken token) {
|
---|
| 156 | var evaluations = 0;
|
---|
| 157 | var N = p1.Solution.Length;
|
---|
| 158 | if (double.IsNaN(p1.Fitness)) {
|
---|
| 159 | Evaluate(p1, token);
|
---|
| 160 | evaluations++;
|
---|
| 161 | }
|
---|
| 162 | if (double.IsNaN(p2.Fitness)) {
|
---|
| 163 | Evaluate(p2, token);
|
---|
| 164 | evaluations++;
|
---|
| 165 | }
|
---|
| 166 | var better = Problem.Maximization ? Math.Max(p1.Fitness, p2.Fitness)
|
---|
| 167 | : Math.Min(p1.Fitness, p2.Fitness);
|
---|
| 168 |
|
---|
| 169 | var offspring = ToScope(null);
|
---|
| 170 | var probe = ToScope(new BinaryVector(N));
|
---|
| 171 | var order = Enumerable.Range(0, N - 1).Shuffle(Context.Random).ToList();
|
---|
| 172 | for (var x = 0; x < N - 1; x++) {
|
---|
| 173 | var b = order[x];
|
---|
| 174 | if (p1.Solution[b] == p2.Solution[b]) continue;
|
---|
| 175 | for (var i = 0; i <= b; i++)
|
---|
| 176 | probe.Solution[i] = p1.Solution[i];
|
---|
| 177 | for (var i = b + 1; i < probe.Solution.Length; i++)
|
---|
| 178 | probe.Solution[i] = p2.Solution[i];
|
---|
| 179 |
|
---|
| 180 | Evaluate(probe, token);
|
---|
| 181 | evaluations++;
|
---|
| 182 | if (Context.IsBetter(probe, offspring)) {
|
---|
| 183 | if (Context.IsBetter(probe.Fitness, better)) {
|
---|
| 184 | Context.IncrementEvaluatedSolutions(evaluations);
|
---|
| 185 | return probe;
|
---|
| 186 | }
|
---|
| 187 | offspring = (ISingleObjectiveSolutionScope<BinaryVector>)probe.Clone();
|
---|
[14420] | 188 | }
|
---|
| 189 | }
|
---|
| 190 |
|
---|
[14544] | 191 | while (evaluations < Context.LocalSearchEvaluations) {
|
---|
| 192 | probe.Solution = UniformCrossover.Apply(Context.Random, p1.Solution, p2.Solution);
|
---|
| 193 | Evaluate(probe, token);
|
---|
| 194 | evaluations++;
|
---|
| 195 | if (Context.IsBetter(probe, offspring)) {
|
---|
| 196 | if (Context.IsBetter(probe.Fitness, better)) {
|
---|
| 197 | Context.IncrementEvaluatedSolutions(evaluations);
|
---|
| 198 | return probe;
|
---|
| 199 | }
|
---|
| 200 | offspring = (ISingleObjectiveSolutionScope<BinaryVector>)probe.Clone();
|
---|
[14420] | 201 | }
|
---|
| 202 | }
|
---|
[14544] | 203 | Context.IncrementEvaluatedSolutions(evaluations);
|
---|
| 204 | return offspring;
|
---|
[14420] | 205 | }
|
---|
| 206 |
|
---|
[14544] | 207 | protected override ISingleObjectiveSolutionScope<BinaryVector> Link(ISingleObjectiveSolutionScope<BinaryVector> a, ISingleObjectiveSolutionScope<BinaryVector> b, CancellationToken token, bool delink = false) {
|
---|
| 208 | var evaluations = 0;
|
---|
[14453] | 209 | if (double.IsNaN(a.Fitness)) {
|
---|
| 210 | Evaluate(a, token);
|
---|
[14544] | 211 | evaluations++;
|
---|
[14453] | 212 | }
|
---|
| 213 | if (double.IsNaN(b.Fitness)) {
|
---|
| 214 | Evaluate(b, token);
|
---|
[14544] | 215 | evaluations++;
|
---|
[14453] | 216 | }
|
---|
[14420] | 217 |
|
---|
[14544] | 218 | var childScope = (ISingleObjectiveSolutionScope<BinaryVector>)a.Clone();
|
---|
[14420] | 219 | var child = childScope.Solution;
|
---|
| 220 | ISingleObjectiveSolutionScope<BinaryVector> best = null;
|
---|
[14544] | 221 | var cF = a.Fitness;
|
---|
[14420] | 222 | var bF = double.NaN;
|
---|
[14544] | 223 | var order = Enumerable.Range(0, child.Length)
|
---|
| 224 | .Where(x => !delink && child[x] != b.Solution[x] || delink && child[x] == b.Solution[x])
|
---|
| 225 | .Shuffle(Context.Random).ToList();
|
---|
| 226 | if (order.Count == 0) return childScope;
|
---|
| 227 |
|
---|
[14420] | 228 | while (true) {
|
---|
| 229 | var bestS = double.NaN;
|
---|
[14544] | 230 | var bestI = -1;
|
---|
| 231 | for (var i = 0; i < order.Count; i++) {
|
---|
[14420] | 232 | var idx = order[i];
|
---|
| 233 | child[idx] = !child[idx]; // move
|
---|
| 234 | Evaluate(childScope, token);
|
---|
[14453] | 235 | evaluations++;
|
---|
[14420] | 236 | var s = childScope.Fitness;
|
---|
| 237 | childScope.Fitness = cF;
|
---|
| 238 | child[idx] = !child[idx]; // undo move
|
---|
[14544] | 239 | if (Context.IsBetter(s, cF)) {
|
---|
[14420] | 240 | bestS = s;
|
---|
[14544] | 241 | bestI = i;
|
---|
[14420] | 242 | break; // first-improvement
|
---|
| 243 | }
|
---|
[14544] | 244 | if (Context.IsBetter(s, bestS)) {
|
---|
[14420] | 245 | // least-degrading
|
---|
| 246 | bestS = s;
|
---|
[14544] | 247 | bestI = i;
|
---|
[14420] | 248 | }
|
---|
| 249 | }
|
---|
[14544] | 250 | child[order[bestI]] = !child[order[bestI]];
|
---|
| 251 | order.RemoveAt(bestI);
|
---|
[14420] | 252 | cF = bestS;
|
---|
| 253 | childScope.Fitness = cF;
|
---|
[14544] | 254 | if (Context.IsBetter(cF, bF)) {
|
---|
[14420] | 255 | bF = cF;
|
---|
| 256 | best = (ISingleObjectiveSolutionScope<BinaryVector>)childScope.Clone();
|
---|
| 257 | }
|
---|
[14544] | 258 | if (order.Count == 0) break;
|
---|
[14420] | 259 | }
|
---|
[14453] | 260 | Context.IncrementEvaluatedSolutions(evaluations);
|
---|
[14420] | 261 | return best ?? childScope;
|
---|
| 262 | }
|
---|
| 263 | }
|
---|
| 264 | }
|
---|