[13412] | 1 | #region License Information
|
---|
| 2 | /* HeuristicLab
|
---|
[16453] | 3 | * Copyright (C) 2002-2019 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
|
---|
[13412] | 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.Linq;
|
---|
| 24 | using HeuristicLab.Common;
|
---|
[12191] | 25 | using HeuristicLab.Core;
|
---|
[13412] | 26 | using HeuristicLab.Data;
|
---|
| 27 | using HeuristicLab.Encodings.PermutationEncoding;
|
---|
[13470] | 28 | using HeuristicLab.Optimization;
|
---|
[13412] | 29 | using HeuristicLab.Parameters;
|
---|
[12191] | 30 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
| 31 | using HeuristicLab.Problems.Instances;
|
---|
[12261] | 32 | using HeuristicLab.Random;
|
---|
[12191] | 33 |
|
---|
| 34 | namespace HeuristicLab.Problems.PTSP {
|
---|
[13412] | 35 | [Item("Estimated Probabilistic Traveling Salesman Problem (PTSP)", "Represents a probabilistic traveling salesman problem where the expected tour length is estimated by averaging over the length of tours on a number of, so called, realizations.")]
|
---|
| 36 | [Creatable(CreatableAttribute.Categories.CombinatorialProblems)]
|
---|
[12191] | 37 | [StorableClass]
|
---|
[13412] | 38 | public sealed class EstimatedProbabilisticTravelingSalesmanProblem : ProbabilisticTravelingSalesmanProblem {
|
---|
[12191] | 39 |
|
---|
[12306] | 40 | #region Parameter Properties
|
---|
[13412] | 41 | public IValueParameter<ItemList<BoolArray>> RealizationsParameter {
|
---|
| 42 | get { return (IValueParameter<ItemList<BoolArray>>)Parameters["Realizations"]; }
|
---|
[12306] | 43 | }
|
---|
[13412] | 44 | public IFixedValueParameter<IntValue> RealizationsSizeParameter {
|
---|
| 45 | get { return (IFixedValueParameter<IntValue>)Parameters["RealizationsSize"]; }
|
---|
[12387] | 46 | }
|
---|
[12306] | 47 | #endregion
|
---|
| 48 |
|
---|
| 49 | #region Properties
|
---|
[13412] | 50 | public ItemList<BoolArray> Realizations {
|
---|
[12306] | 51 | get { return RealizationsParameter.Value; }
|
---|
| 52 | set { RealizationsParameter.Value = value; }
|
---|
| 53 | }
|
---|
[12387] | 54 |
|
---|
[13412] | 55 | public int RealizationsSize {
|
---|
| 56 | get { return RealizationsSizeParameter.Value.Value; }
|
---|
| 57 | set { RealizationsSizeParameter.Value.Value = value; }
|
---|
[12387] | 58 | }
|
---|
[12306] | 59 | #endregion
|
---|
| 60 |
|
---|
[12191] | 61 | [StorableConstructor]
|
---|
| 62 | private EstimatedProbabilisticTravelingSalesmanProblem(bool deserializing) : base(deserializing) { }
|
---|
| 63 | private EstimatedProbabilisticTravelingSalesmanProblem(EstimatedProbabilisticTravelingSalesmanProblem original, Cloner cloner)
|
---|
| 64 | : base(original, cloner) {
|
---|
[12387] | 65 | RegisterEventHandlers();
|
---|
[12191] | 66 | }
|
---|
[12387] | 67 | public EstimatedProbabilisticTravelingSalesmanProblem() {
|
---|
[13412] | 68 | Parameters.Add(new FixedValueParameter<IntValue>("RealizationsSize", "Size of the sample for the estimation-based evaluation", new IntValue(100)));
|
---|
| 69 | Parameters.Add(new ValueParameter<ItemList<BoolArray>>("Realizations", "The list of samples drawn from all possible stochastic instances.", new ItemList<BoolArray>()));
|
---|
[12799] | 70 |
|
---|
| 71 | Operators.Add(new BestPTSPSolutionAnalyzer());
|
---|
| 72 |
|
---|
[13412] | 73 | Operators.Add(new PTSPEstimatedInversionMoveEvaluator());
|
---|
| 74 | Operators.Add(new PTSPEstimatedInsertionMoveEvaluator());
|
---|
[13470] | 75 | Operators.Add(new PTSPEstimatedInversionLocalImprovement());
|
---|
| 76 | Operators.Add(new PTSPEstimatedInsertionLocalImprovement());
|
---|
| 77 | Operators.Add(new PTSPEstimatedTwoPointFiveLocalImprovement());
|
---|
[12191] | 78 |
|
---|
[13412] | 79 | Operators.Add(new ExhaustiveTwoPointFiveMoveGenerator());
|
---|
| 80 | Operators.Add(new StochasticTwoPointFiveMultiMoveGenerator());
|
---|
| 81 | Operators.Add(new StochasticTwoPointFiveSingleMoveGenerator());
|
---|
[12387] | 82 | Operators.Add(new TwoPointFiveMoveMaker());
|
---|
[13470] | 83 | Operators.Add(new PTSPEstimatedTwoPointFiveMoveEvaluator());
|
---|
[12387] | 84 |
|
---|
[13470] | 85 | Operators.RemoveAll(x => x is SingleObjectiveMoveGenerator);
|
---|
| 86 | Operators.RemoveAll(x => x is SingleObjectiveMoveMaker);
|
---|
| 87 | Operators.RemoveAll(x => x is SingleObjectiveMoveEvaluator);
|
---|
| 88 |
|
---|
[12387] | 89 | Encoding.ConfigureOperators(Operators.OfType<IOperator>());
|
---|
[13412] | 90 | foreach (var twopointfiveMoveOperator in Operators.OfType<ITwoPointFiveMoveOperator>()) {
|
---|
[12387] | 91 | twopointfiveMoveOperator.TwoPointFiveMoveParameter.ActualName = "Permutation.TwoPointFiveMove";
|
---|
| 92 | }
|
---|
[13412] | 93 |
|
---|
[13470] | 94 | UpdateRealizations();
|
---|
[12387] | 95 | RegisterEventHandlers();
|
---|
| 96 | }
|
---|
| 97 |
|
---|
[12191] | 98 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
| 99 | return new EstimatedProbabilisticTravelingSalesmanProblem(this, cloner);
|
---|
| 100 | }
|
---|
| 101 |
|
---|
[13412] | 102 | [StorableHook(HookType.AfterDeserialization)]
|
---|
| 103 | private void AfterDeserialization() {
|
---|
| 104 | RegisterEventHandlers();
|
---|
| 105 | }
|
---|
| 106 |
|
---|
| 107 | private void RegisterEventHandlers() {
|
---|
| 108 | RealizationsSizeParameter.Value.ValueChanged += RealizationsSizeParameter_ValueChanged;
|
---|
| 109 | }
|
---|
| 110 |
|
---|
| 111 | private void RealizationsSizeParameter_ValueChanged(object sender, EventArgs e) {
|
---|
| 112 | UpdateRealizations();
|
---|
| 113 | }
|
---|
| 114 |
|
---|
| 115 | public override double Evaluate(Permutation tour, IRandom random) {
|
---|
[13470] | 116 | // abeham: Cache parameters in local variables for performance reasons
|
---|
| 117 | var realizations = Realizations;
|
---|
| 118 | var realizationsSize = RealizationsSize;
|
---|
| 119 | var useDistanceMatrix = UseDistanceMatrix;
|
---|
| 120 | var distanceMatrix = DistanceMatrix;
|
---|
| 121 | var distanceCalculator = DistanceCalculator;
|
---|
| 122 | var coordinates = Coordinates;
|
---|
| 123 |
|
---|
| 124 | // Estimation-based evaluation, here without calculating variance for faster evaluation
|
---|
[13412] | 125 | var estimatedSum = 0.0;
|
---|
[13470] | 126 | for (var i = 0; i < realizations.Count; i++) {
|
---|
[12261] | 127 | int singleRealization = -1, firstNode = -1;
|
---|
[13470] | 128 | for (var j = 0; j < realizations[i].Length; j++) {
|
---|
| 129 | if (realizations[i][tour[j]]) {
|
---|
[12191] | 130 | if (singleRealization != -1) {
|
---|
[13470] | 131 | estimatedSum += useDistanceMatrix ? distanceMatrix[singleRealization, tour[j]] : distanceCalculator.Calculate(singleRealization, tour[j], coordinates);
|
---|
[12261] | 132 | } else {
|
---|
[13412] | 133 | firstNode = tour[j];
|
---|
[12191] | 134 | }
|
---|
[13412] | 135 | singleRealization = tour[j];
|
---|
[12191] | 136 | }
|
---|
| 137 | }
|
---|
[12261] | 138 | if (singleRealization != -1) {
|
---|
[13470] | 139 | estimatedSum += useDistanceMatrix ? distanceMatrix[singleRealization, firstNode] : distanceCalculator.Calculate(singleRealization, firstNode, coordinates);
|
---|
[12261] | 140 | }
|
---|
[12191] | 141 | }
|
---|
[13470] | 142 | return estimatedSum / realizationsSize;
|
---|
[12191] | 143 | }
|
---|
| 144 |
|
---|
[13412] | 145 | /// <summary>
|
---|
| 146 | /// An evaluate method that can be used if mean as well as variance should be calculated
|
---|
| 147 | /// </summary>
|
---|
| 148 | /// <param name="tour">The tour between all cities.</param>
|
---|
| 149 | /// <param name="distanceMatrix">The distances between the cities.</param>
|
---|
| 150 | /// <param name="realizations">A sample of realizations of the stochastic instance</param>
|
---|
[13470] | 151 | /// <param name="variance">The estimated variance will be returned in addition to the mean.</param>
|
---|
[13412] | 152 | /// <returns>A vector with length two containing mean and variance.</returns>
|
---|
[13470] | 153 | public static double Evaluate(Permutation tour, DistanceMatrix distanceMatrix, ItemList<BoolArray> realizations, out double variance) {
|
---|
| 154 | return Evaluate(tour, (a, b) => distanceMatrix[a, b], realizations, out variance);
|
---|
| 155 | }
|
---|
| 156 |
|
---|
| 157 | /// <summary>
|
---|
| 158 | /// An evaluate method that can be used if mean as well as variance should be calculated
|
---|
| 159 | /// </summary>
|
---|
| 160 | /// <param name="tour">The tour between all cities.</param>
|
---|
| 161 | /// <param name="distance">A func that accepts the index of two cities and returns the distance as a double.</param>
|
---|
| 162 | /// <param name="realizations">A sample of realizations of the stochastic instance</param>
|
---|
| 163 | /// <param name="variance">The estimated variance will be returned in addition to the mean.</param>
|
---|
| 164 | /// <returns>A vector with length two containing mean and variance.</returns>
|
---|
| 165 | public static double Evaluate(Permutation tour, Func<int, int, double> distance, ItemList<BoolArray> realizations, out double variance) {
|
---|
[12261] | 166 | // Estimation-based evaluation
|
---|
[13412] | 167 | var estimatedSum = 0.0;
|
---|
| 168 | var partialSums = new double[realizations.Count];
|
---|
| 169 | for (var i = 0; i < realizations.Count; i++) {
|
---|
[12261] | 170 | partialSums[i] = 0;
|
---|
| 171 | int singleRealization = -1, firstNode = -1;
|
---|
[13412] | 172 | for (var j = 0; j < realizations[i].Length; j++) {
|
---|
| 173 | if (realizations[i][tour[j]]) {
|
---|
[12261] | 174 | if (singleRealization != -1) {
|
---|
[13470] | 175 | partialSums[i] += distance(singleRealization, tour[j]);
|
---|
[12261] | 176 | } else {
|
---|
[13412] | 177 | firstNode = tour[j];
|
---|
[12261] | 178 | }
|
---|
[13412] | 179 | singleRealization = tour[j];
|
---|
[12261] | 180 | }
|
---|
| 181 | }
|
---|
| 182 | if (singleRealization != -1) {
|
---|
[13470] | 183 | partialSums[i] += distance(singleRealization, firstNode);
|
---|
[12261] | 184 | }
|
---|
| 185 | estimatedSum += partialSums[i];
|
---|
| 186 | }
|
---|
[13412] | 187 | var mean = estimatedSum / realizations.Count;
|
---|
[13470] | 188 | variance = 0.0;
|
---|
[13412] | 189 | for (var i = 0; i < realizations.Count; i++) {
|
---|
[12261] | 190 | variance += Math.Pow((partialSums[i] - mean), 2);
|
---|
| 191 | }
|
---|
| 192 | variance = variance / realizations.Count;
|
---|
[13470] | 193 | return mean;
|
---|
[12261] | 194 | }
|
---|
| 195 |
|
---|
[13470] | 196 | public override void Load(PTSPData data) {
|
---|
| 197 | base.Load(data);
|
---|
| 198 | UpdateRealizations();
|
---|
| 199 |
|
---|
| 200 | foreach (var op in Operators.OfType<IEstimatedPTSPOperator>()) {
|
---|
| 201 | op.RealizationsParameter.ActualName = RealizationsParameter.Name;
|
---|
| 202 | }
|
---|
[12387] | 203 | }
|
---|
[12272] | 204 |
|
---|
[12387] | 205 | private void UpdateRealizations() {
|
---|
[13412] | 206 | var realizations = new ItemList<BoolArray>(RealizationsSize);
|
---|
| 207 | var rand = new MersenneTwister();
|
---|
| 208 | for (var i = 0; i < RealizationsSize; i++) {
|
---|
| 209 | var newRealization = new BoolArray(Probabilities.Length);
|
---|
| 210 | var countOnes = 0;
|
---|
[13470] | 211 | do {
|
---|
[12261] | 212 | countOnes = 0;
|
---|
[13412] | 213 | for (var j = 0; j < Probabilities.Length; j++) {
|
---|
| 214 | newRealization[j] = Probabilities[j] < rand.NextDouble();
|
---|
| 215 | if (newRealization[j]) countOnes++;
|
---|
[12219] | 216 | }
|
---|
[13470] | 217 | // only generate realizations with at least 4 cities visited
|
---|
| 218 | } while (countOnes < 4 && Probabilities.Length > 3);
|
---|
| 219 | realizations.Add(newRealization);
|
---|
[12219] | 220 | }
|
---|
[13412] | 221 | Realizations = realizations;
|
---|
[12387] | 222 | }
|
---|
[12191] | 223 | }
|
---|
| 224 | } |
---|