1 | #region License Information
|
---|
2 | /* HeuristicLab
|
---|
3 | * Copyright (C) 2002-2019 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 HeuristicLab.Common;
|
---|
23 | using HeuristicLab.Core;
|
---|
24 | using System;
|
---|
25 | using System.Collections.Generic;
|
---|
26 | using System.Linq;
|
---|
27 |
|
---|
28 | namespace HeuristicLab.Algorithms.DynamicALPS
|
---|
29 | {
|
---|
30 | public static class DynamicALPSUtil
|
---|
31 | {
|
---|
32 | public static void QuickSort(double[] array, int[] idx, int from, int to)
|
---|
33 | {
|
---|
34 | if (from < to)
|
---|
35 | {
|
---|
36 | double temp = array[to];
|
---|
37 | int tempIdx = idx[to];
|
---|
38 | int i = from - 1;
|
---|
39 | for (int j = from; j < to; j++)
|
---|
40 | {
|
---|
41 | if (array[j] <= temp)
|
---|
42 | {
|
---|
43 | i++;
|
---|
44 | double tempValue = array[j];
|
---|
45 | array[j] = array[i];
|
---|
46 | array[i] = tempValue;
|
---|
47 | int tempIndex = idx[j];
|
---|
48 | idx[j] = idx[i];
|
---|
49 | idx[i] = tempIndex;
|
---|
50 | }
|
---|
51 | }
|
---|
52 | array[to] = array[i + 1];
|
---|
53 | array[i + 1] = temp;
|
---|
54 | idx[to] = idx[i + 1];
|
---|
55 | idx[i + 1] = tempIdx;
|
---|
56 | QuickSort(array, idx, from, i);
|
---|
57 | QuickSort(array, idx, i + 1, to);
|
---|
58 | }
|
---|
59 | }
|
---|
60 |
|
---|
61 | public static void MinFastSort(double[] x, int[] idx, int n, int m)
|
---|
62 | {
|
---|
63 | for (int i = 0; i < m; i++)
|
---|
64 | {
|
---|
65 | for (int j = i + 1; j < n; j++)
|
---|
66 | {
|
---|
67 | if (x[i] > x[j])
|
---|
68 | {
|
---|
69 | double temp = x[i];
|
---|
70 | x[i] = x[j];
|
---|
71 | x[j] = temp;
|
---|
72 | int id = idx[i];
|
---|
73 | idx[i] = idx[j];
|
---|
74 | idx[j] = id;
|
---|
75 | }
|
---|
76 | }
|
---|
77 | }
|
---|
78 | }
|
---|
79 |
|
---|
80 | public static IList<IDynamicALPSSolution> GetSubsetOfEvenlyDistributedSolutions(IRandom random, IList<IDynamicALPSSolution> solutionList, int newSolutionListSize)
|
---|
81 | {
|
---|
82 | if (solutionList == null || solutionList.Count == 0)
|
---|
83 | {
|
---|
84 | throw new ArgumentException("Solution list is null or empty.");
|
---|
85 | }
|
---|
86 |
|
---|
87 | return solutionList[0].Dimensions == 2
|
---|
88 | ? TwoObjectivesCase(solutionList, newSolutionListSize)
|
---|
89 | : MoreThanTwoObjectivesCase(random, solutionList, newSolutionListSize);
|
---|
90 | }
|
---|
91 |
|
---|
92 | private static IList<IDynamicALPSSolution> TwoObjectivesCase(IList<IDynamicALPSSolution> solutionList, int newSolutionListSize)
|
---|
93 | {
|
---|
94 | var resultSolutionList = new IDynamicALPSSolution[newSolutionListSize];
|
---|
95 |
|
---|
96 | // compute weight vectors
|
---|
97 | double[][] lambda_moead = new double[newSolutionListSize][];
|
---|
98 | var values = SequenceGenerator.GenerateSteps(0m, 1m, 1m / newSolutionListSize).ToArray();
|
---|
99 | for (int i = 0; i < newSolutionListSize; ++i)
|
---|
100 | {
|
---|
101 | var weights = new double[newSolutionListSize];
|
---|
102 | weights[0] = (double)values[i];
|
---|
103 | weights[1] = 1 - weights[0];
|
---|
104 |
|
---|
105 | lambda_moead[i] = weights;
|
---|
106 | }
|
---|
107 |
|
---|
108 | var idealPoint = new double[2];
|
---|
109 | foreach (var solution in solutionList)
|
---|
110 | {
|
---|
111 | // update ideal point
|
---|
112 | idealPoint.UpdateIdeal(solution.Qualities);
|
---|
113 | }
|
---|
114 |
|
---|
115 | // Select the best solution for each weight vector
|
---|
116 | for (int i = 0; i < newSolutionListSize; i++)
|
---|
117 | {
|
---|
118 | var currentBest = solutionList[0];
|
---|
119 | double value = ScalarizingFitnessFunction(currentBest, lambda_moead[i], idealPoint);
|
---|
120 | for (int j = 1; j < solutionList.Count; j++)
|
---|
121 | {
|
---|
122 | double aux = ScalarizingFitnessFunction(solutionList[j], lambda_moead[i], idealPoint); // we are looking for the best for the weight i
|
---|
123 | if (aux < value)
|
---|
124 | { // solution in position j is better!
|
---|
125 | value = aux;
|
---|
126 | currentBest = solutionList[j];
|
---|
127 | }
|
---|
128 | }
|
---|
129 | resultSolutionList[i] = (DynamicALPSSolution)currentBest.Clone();
|
---|
130 | }
|
---|
131 |
|
---|
132 | return resultSolutionList;
|
---|
133 | }
|
---|
134 |
|
---|
135 | private static IList<IDynamicALPSSolution> MoreThanTwoObjectivesCase(IRandom random, IList<IDynamicALPSSolution> solutionList, int newSolutionListSize)
|
---|
136 | {
|
---|
137 | var resultSolutionList = new List<IDynamicALPSSolution>();
|
---|
138 |
|
---|
139 | int randomIndex = random.Next(0, solutionList.Count);
|
---|
140 |
|
---|
141 | var candidate = new List<IDynamicALPSSolution>();
|
---|
142 | resultSolutionList.Add(solutionList[randomIndex]);
|
---|
143 |
|
---|
144 | for (int i = 0; i < solutionList.Count; ++i)
|
---|
145 | {
|
---|
146 | if (i != randomIndex)
|
---|
147 | {
|
---|
148 | candidate.Add(solutionList[i]);
|
---|
149 | }
|
---|
150 | }
|
---|
151 |
|
---|
152 | while (resultSolutionList.Count < newSolutionListSize)
|
---|
153 | {
|
---|
154 | int index = 0;
|
---|
155 | var selected = candidate[0]; // it should be a next! (n <= population size!)
|
---|
156 | double aux = CalculateBestDistance(selected, solutionList);
|
---|
157 | int i = 1;
|
---|
158 | while (i < candidate.Count)
|
---|
159 | {
|
---|
160 | var nextCandidate = candidate[i];
|
---|
161 | double distanceValue = CalculateBestDistance(nextCandidate, solutionList);
|
---|
162 | if (aux < distanceValue)
|
---|
163 | {
|
---|
164 | index = i;
|
---|
165 | aux = distanceValue;
|
---|
166 | }
|
---|
167 | i++;
|
---|
168 | }
|
---|
169 |
|
---|
170 | // add the selected to res and remove from candidate list
|
---|
171 | var removedSolution = candidate[index];
|
---|
172 | candidate.RemoveAt(index);
|
---|
173 | resultSolutionList.Add((DynamicALPSSolution)removedSolution.Clone());
|
---|
174 | }
|
---|
175 |
|
---|
176 | return resultSolutionList;
|
---|
177 | }
|
---|
178 |
|
---|
179 | private static double ScalarizingFitnessFunction(IDynamicALPSSolution currentBest, double[] lambda_moead, double[] idealPoint)
|
---|
180 | {
|
---|
181 | double maxFun = -1.0e+30;
|
---|
182 |
|
---|
183 | for (int n = 0; n < idealPoint.Length; n++)
|
---|
184 | {
|
---|
185 | double diff = Math.Abs(currentBest.Qualities[n] - idealPoint[n]);
|
---|
186 |
|
---|
187 | double functionValue;
|
---|
188 | if (lambda_moead[n] == 0)
|
---|
189 | {
|
---|
190 | functionValue = 0.0001 * diff;
|
---|
191 | }
|
---|
192 | else
|
---|
193 | {
|
---|
194 | functionValue = diff * lambda_moead[n];
|
---|
195 | }
|
---|
196 | if (functionValue > maxFun)
|
---|
197 | {
|
---|
198 | maxFun = functionValue;
|
---|
199 | }
|
---|
200 | }
|
---|
201 |
|
---|
202 | return maxFun;
|
---|
203 | }
|
---|
204 |
|
---|
205 | public static void UpdateIdeal(this double[] idealPoint, double[] point)
|
---|
206 | {
|
---|
207 | for (int i = 0; i < point.Length; ++i)
|
---|
208 | {
|
---|
209 | if (double.IsInfinity(point[i]) || double.IsNaN(point[i]))
|
---|
210 | {
|
---|
211 | continue;
|
---|
212 | }
|
---|
213 |
|
---|
214 | if (idealPoint[i] > point[i])
|
---|
215 | {
|
---|
216 | idealPoint[i] = point[i];
|
---|
217 | }
|
---|
218 | }
|
---|
219 | }
|
---|
220 |
|
---|
221 | public static void UpdateNadir(this double[] nadirPoint, double[] point)
|
---|
222 | {
|
---|
223 | for (int i = 0; i < point.Length; ++i)
|
---|
224 | {
|
---|
225 | if (double.IsInfinity(point[i]) || double.IsNaN(point[i]))
|
---|
226 | {
|
---|
227 | continue;
|
---|
228 | }
|
---|
229 |
|
---|
230 | if (nadirPoint[i] < point[i])
|
---|
231 | {
|
---|
232 | nadirPoint[i] = point[i];
|
---|
233 | }
|
---|
234 | }
|
---|
235 | }
|
---|
236 |
|
---|
237 | public static void UpdateIdeal(this double[] idealPoint, IList<IDynamicALPSSolution> solutions)
|
---|
238 | {
|
---|
239 | foreach (var s in solutions)
|
---|
240 | {
|
---|
241 | idealPoint.UpdateIdeal(s.Qualities);
|
---|
242 | }
|
---|
243 | }
|
---|
244 |
|
---|
245 | public static void UpdateNadir(this double[] nadirPoint, IList<IDynamicALPSSolution> solutions)
|
---|
246 | {
|
---|
247 | foreach (var s in solutions)
|
---|
248 | {
|
---|
249 | nadirPoint.UpdateNadir(s.Qualities);
|
---|
250 | }
|
---|
251 | }
|
---|
252 |
|
---|
253 | private static double CalculateBestDistance(IDynamicALPSSolution solution, IList<IDynamicALPSSolution> solutionList)
|
---|
254 | {
|
---|
255 | var best = solutionList.Min(x => EuclideanDistance(solution.Qualities, x.Qualities));
|
---|
256 | if (double.IsNaN(best) || double.IsInfinity(best))
|
---|
257 | {
|
---|
258 | best = double.MaxValue;
|
---|
259 | }
|
---|
260 | return best;
|
---|
261 | }
|
---|
262 |
|
---|
263 | public static double EuclideanDistance(double[] a, double[] b)
|
---|
264 | {
|
---|
265 | if (a.Length != b.Length)
|
---|
266 | {
|
---|
267 | throw new ArgumentException("Euclidean distance: the arrays have different lengths.");
|
---|
268 | }
|
---|
269 |
|
---|
270 | var distance = 0d;
|
---|
271 | for (int i = 0; i < a.Length; ++i)
|
---|
272 | {
|
---|
273 | var d = a[i] - b[i];
|
---|
274 | distance += d * d;
|
---|
275 | }
|
---|
276 | return Math.Sqrt(distance);
|
---|
277 | }
|
---|
278 | }
|
---|
279 | }
|
---|