1 | #region License Information
|
---|
2 | /* HeuristicLab
|
---|
3 | * Copyright (C) 2002-2012 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 HeuristicLab.Common;
|
---|
24 | using HeuristicLab.Core;
|
---|
25 | using HeuristicLab.Data;
|
---|
26 | using HeuristicLab.Encodings.IntegerVectorEncoding;
|
---|
27 | using HeuristicLab.Parameters;
|
---|
28 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
29 |
|
---|
30 | namespace HeuristicLab.Problems.GeneralizedQuadraticAssignment {
|
---|
31 | [Item("CordeauCrossover", "The merge procedure that is described in Cordeau, J.-F., Gaudioso, M., Laporte, G., Moccia, L. 2006. A memetic heuristic for the generalized quadratic assignment problem. INFORMS Journal on Computing, 18, pp. 433–443.")]
|
---|
32 | [StorableClass]
|
---|
33 | public class CordeauCrossover : GQAPCrossover,
|
---|
34 | IQualitiesAwareGQAPOperator, IWeightsAwareGQAPOperator, IDistancesAwareGQAPOperator, IInstallationCostsAwareGQAPOperator,
|
---|
35 | IDemandsAwareGQAPOperator, ICapacitiesAwareGQAPOperator, ITransportationCostsAwareGQAPOperator,
|
---|
36 | IOverbookedCapacityPenaltyAwareGQAPOperator {
|
---|
37 |
|
---|
38 | public ILookupParameter<BoolValue> MaximizationParameter {
|
---|
39 | get { return (ILookupParameter<BoolValue>)Parameters["Maximization"]; }
|
---|
40 | }
|
---|
41 | public IScopeTreeLookupParameter<DoubleValue> QualityParameter {
|
---|
42 | get { return (IScopeTreeLookupParameter<DoubleValue>)Parameters["Quality"]; }
|
---|
43 | }
|
---|
44 | public IScopeTreeLookupParameter<DoubleValue> FlowDistanceQualityParameter {
|
---|
45 | get { return (IScopeTreeLookupParameter<DoubleValue>)Parameters["FlowDistanceQuality"]; }
|
---|
46 | }
|
---|
47 | public IScopeTreeLookupParameter<DoubleValue> InstallationQualityParameter {
|
---|
48 | get { return (IScopeTreeLookupParameter<DoubleValue>)Parameters["InstallationQuality"]; }
|
---|
49 | }
|
---|
50 | public IScopeTreeLookupParameter<DoubleValue> OverbookedCapacityParameter {
|
---|
51 | get { return (IScopeTreeLookupParameter<DoubleValue>)Parameters["OverbookedCapacity"]; }
|
---|
52 | }
|
---|
53 | public ILookupParameter<DoubleMatrix> WeightsParameter {
|
---|
54 | get { return (ILookupParameter<DoubleMatrix>)Parameters["Weights"]; }
|
---|
55 | }
|
---|
56 | public ILookupParameter<DoubleMatrix> DistancesParameter {
|
---|
57 | get { return (ILookupParameter<DoubleMatrix>)Parameters["Distances"]; }
|
---|
58 | }
|
---|
59 | public ILookupParameter<DoubleMatrix> InstallationCostsParameter {
|
---|
60 | get { return (ILookupParameter<DoubleMatrix>)Parameters["InstallationCosts"]; }
|
---|
61 | }
|
---|
62 | public ILookupParameter<DoubleArray> DemandsParameter {
|
---|
63 | get { return (ILookupParameter<DoubleArray>)Parameters["Demands"]; }
|
---|
64 | }
|
---|
65 | public ILookupParameter<DoubleArray> CapacitiesParameter {
|
---|
66 | get { return (ILookupParameter<DoubleArray>)Parameters["Capacities"]; }
|
---|
67 | }
|
---|
68 | public IValueLookupParameter<DoubleValue> TransportationCostsParameter {
|
---|
69 | get { return (IValueLookupParameter<DoubleValue>)Parameters["TransportationCosts"]; }
|
---|
70 | }
|
---|
71 | public IValueLookupParameter<DoubleValue> OverbookedCapacityPenaltyParameter {
|
---|
72 | get { return (IValueLookupParameter<DoubleValue>)Parameters["OverbookedCapacityPenalty"]; }
|
---|
73 | }
|
---|
74 |
|
---|
75 | [StorableConstructor]
|
---|
76 | protected CordeauCrossover(bool deserializing) : base(deserializing) { }
|
---|
77 | protected CordeauCrossover(CordeauCrossover original, Cloner cloner)
|
---|
78 | : base(original, cloner) {
|
---|
79 | }
|
---|
80 | public CordeauCrossover()
|
---|
81 | : base() {
|
---|
82 | Parameters.Add(new LookupParameter<BoolValue>("Maximization", GeneralizedQuadraticAssignmentProblem.MaximizationDescription));
|
---|
83 | Parameters.Add(new ScopeTreeLookupParameter<DoubleValue>("Quality", GQAPEvaluator.QualityDescription, 1));
|
---|
84 | Parameters.Add(new ScopeTreeLookupParameter<DoubleValue>("FlowDistanceQuality", GQAPEvaluator.FlowDistanceQualityDescription, 1));
|
---|
85 | Parameters.Add(new ScopeTreeLookupParameter<DoubleValue>("InstallationQuality", GQAPEvaluator.InstallationQualityDescription, 1));
|
---|
86 | Parameters.Add(new ScopeTreeLookupParameter<DoubleValue>("OverbookedCapacity", GQAPEvaluator.OverbookedCapacityDescription, 1));
|
---|
87 | Parameters.Add(new LookupParameter<DoubleMatrix>("Weights", GeneralizedQuadraticAssignmentProblem.WeightsDescription));
|
---|
88 | Parameters.Add(new LookupParameter<DoubleMatrix>("Distances", GeneralizedQuadraticAssignmentProblem.DistancesDescription));
|
---|
89 | Parameters.Add(new LookupParameter<DoubleMatrix>("InstallationCosts", GeneralizedQuadraticAssignmentProblem.InstallationCostsDescription));
|
---|
90 | Parameters.Add(new LookupParameter<DoubleArray>("Demands", GeneralizedQuadraticAssignmentProblem.DemandsDescription));
|
---|
91 | Parameters.Add(new LookupParameter<DoubleArray>("Capacities", GeneralizedQuadraticAssignmentProblem.CapacitiesDescription));
|
---|
92 | Parameters.Add(new ValueLookupParameter<DoubleValue>("TransportationCosts", GeneralizedQuadraticAssignmentProblem.TransportationCostsDescription));
|
---|
93 | Parameters.Add(new ValueLookupParameter<DoubleValue>("OverbookedCapacityPenalty", GeneralizedQuadraticAssignmentProblem.OverbookedCapacityPenaltyDescription));
|
---|
94 | }
|
---|
95 |
|
---|
96 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
97 | return new CordeauCrossover(this, cloner);
|
---|
98 | }
|
---|
99 |
|
---|
100 | public static IntegerVector Apply(IRandom random, BoolValue maximization,
|
---|
101 | IntegerVector parent1, DoubleValue quality1,
|
---|
102 | IntegerVector parent2, DoubleValue quality2,
|
---|
103 | DoubleMatrix weights, DoubleMatrix distances, DoubleMatrix installationCosts,
|
---|
104 | DoubleArray demands, DoubleArray capacities,
|
---|
105 | DoubleValue transportationCosts, DoubleValue overbookedCapacityPenalty) {
|
---|
106 | var mediana = Inizialize(distances);
|
---|
107 | int m = capacities.Length;
|
---|
108 | int n = demands.Length;
|
---|
109 | int i, j, k, ik1, ik2;
|
---|
110 | int control;
|
---|
111 | bool nofound = false, onefound = false;
|
---|
112 | double fbest;
|
---|
113 | IntegerVector son = new IntegerVector(parent1.Length), result = null;
|
---|
114 |
|
---|
115 | fbest = quality1.Value;
|
---|
116 | if (maximization.Value && quality1.Value < quality2.Value
|
---|
117 | || !maximization.Value && quality1.Value > quality2.Value) {
|
---|
118 | var temp = parent1;
|
---|
119 | parent1 = parent2;
|
---|
120 | parent2 = temp;
|
---|
121 | fbest = quality2.Value;
|
---|
122 | }
|
---|
123 | var cap = new double[m];
|
---|
124 | for (i = 0; i < m; i++) {
|
---|
125 |
|
---|
126 | for (j = 0; j < m; j++) cap[j] = 0;
|
---|
127 |
|
---|
128 | for (k = 0; k < n; k++) {
|
---|
129 | son[k] = -1;
|
---|
130 | ik1 = parent1[k];
|
---|
131 | ik2 = parent2[k];
|
---|
132 | if (distances[i, ik1] < mediana[i]) {
|
---|
133 | son[k] = ik1;
|
---|
134 | cap[ik1] += demands[k];
|
---|
135 | } else if (distances[i, ik2] > mediana[i]) {
|
---|
136 | son[k] = ik2;
|
---|
137 | cap[ik2] += demands[k];
|
---|
138 | }
|
---|
139 | }
|
---|
140 | k = 0;
|
---|
141 | nofound = false;
|
---|
142 | while (k < n && !nofound) {
|
---|
143 | if (son[k] < 0) {
|
---|
144 | control = 0;
|
---|
145 | do {
|
---|
146 | j = random.Next(m);
|
---|
147 | control++;
|
---|
148 | }
|
---|
149 | while (cap[j] + demands[k] > capacities[j] && control < 3 * m);
|
---|
150 | if (cap[j] + demands[k] <= capacities[j]) {
|
---|
151 | son[k] = j;
|
---|
152 | cap[j] += demands[k];
|
---|
153 | } else {
|
---|
154 | nofound = true;
|
---|
155 | }
|
---|
156 | }
|
---|
157 | k++;
|
---|
158 | }
|
---|
159 | if (!nofound) {
|
---|
160 | double sonQual = GQAPEvaluator.Evaluate(son, weights, distances, installationCosts, demands, capacities, transportationCosts, overbookedCapacityPenalty);
|
---|
161 | if (sonQual < fbest) {
|
---|
162 | fbest = sonQual;
|
---|
163 | result = (IntegerVector)son.Clone();
|
---|
164 | onefound = true;
|
---|
165 | }
|
---|
166 | }/* else {
|
---|
167 | abortedmerge++;
|
---|
168 | printf("aborted merge %8d\n", abortedmerge);
|
---|
169 | } */
|
---|
170 | }
|
---|
171 | if (!onefound && !nofound) {
|
---|
172 | i = random.Next(m);
|
---|
173 | for (j = 0; j < m; j++) {
|
---|
174 | cap[j] = 0.0;
|
---|
175 | }
|
---|
176 | for (k = 0; k < n; k++) {
|
---|
177 | son[k] = -1;
|
---|
178 | ik1 = parent1[k];
|
---|
179 | ik2 = parent2[k];
|
---|
180 | if (distances[i, ik1] < mediana[i]) {
|
---|
181 | son[k] = ik1;
|
---|
182 | cap[ik1] += demands[k];
|
---|
183 | } else if (distances[i, ik2] > mediana[i]) {
|
---|
184 | son[k] = ik2;
|
---|
185 | cap[ik2] += demands[k];
|
---|
186 | }
|
---|
187 | }
|
---|
188 | for (k = 0; k < n; k++) {
|
---|
189 | if (son[k] < 0) {
|
---|
190 | j = random.Next(m);
|
---|
191 | son[k] = j;
|
---|
192 | cap[j] += demands[k];
|
---|
193 | }
|
---|
194 | }
|
---|
195 | if (result == null) {
|
---|
196 | result = (IntegerVector)son.Clone();
|
---|
197 | onefound = true;
|
---|
198 | } else {
|
---|
199 | double sonQual = GQAPEvaluator.Evaluate(son, weights, distances, installationCosts, demands, capacities, transportationCosts, overbookedCapacityPenalty);
|
---|
200 | if (sonQual < fbest) {
|
---|
201 | fbest = sonQual;
|
---|
202 | result = (IntegerVector)son.Clone();
|
---|
203 | onefound = true;
|
---|
204 | }
|
---|
205 | }
|
---|
206 | /*if (tabufix(&son, 0.5 * sqrt(n * m), round(n * m * log10(n)), &tabufix_it)) {
|
---|
207 | solution_cost(&son);
|
---|
208 | if (son.cost < fbest) {
|
---|
209 | fbest = son.cost;
|
---|
210 | *sptr = son;
|
---|
211 | onefound = TRUE;
|
---|
212 | merge_fixed++;
|
---|
213 | }*/
|
---|
214 | }
|
---|
215 | if (result == null) throw new InvalidOperationException("Child could not be created.");
|
---|
216 | return result;
|
---|
217 | }
|
---|
218 |
|
---|
219 | protected override IntegerVector Cross(IRandom random, ItemArray<IntegerVector> parents) {
|
---|
220 | if (parents == null) throw new ArgumentNullException("parents");
|
---|
221 | if (parents.Length != 2) throw new ArgumentException(Name + " works only with exactly two parents.");
|
---|
222 |
|
---|
223 | var qualities = QualityParameter.ActualValue;
|
---|
224 | return Apply(random, MaximizationParameter.ActualValue,
|
---|
225 | parents[0], qualities[0],
|
---|
226 | parents[1], qualities[1],
|
---|
227 | WeightsParameter.ActualValue, DistancesParameter.ActualValue, InstallationCostsParameter.ActualValue,
|
---|
228 | DemandsParameter.ActualValue, CapacitiesParameter.ActualValue,
|
---|
229 | TransportationCostsParameter.ActualValue, OverbookedCapacityPenaltyParameter.ActualValue);
|
---|
230 | }
|
---|
231 |
|
---|
232 | private static double[] Inizialize(DoubleMatrix distances) {
|
---|
233 | int i, j;
|
---|
234 | double mdi;
|
---|
235 | double delta;
|
---|
236 | int ilow, ihigh;
|
---|
237 | bool balance;
|
---|
238 | int operation;
|
---|
239 | int count;
|
---|
240 | int m = distances.Rows;
|
---|
241 |
|
---|
242 | double[] mediana = new double[distances.Rows];
|
---|
243 |
|
---|
244 | for (i = 0; i < m; i++) {
|
---|
245 | mdi = 0;
|
---|
246 | for (j = 0; j < m; j++) {
|
---|
247 | mdi += distances[i, j];
|
---|
248 | }
|
---|
249 | mediana[i] = mdi / m;
|
---|
250 | balance = false;
|
---|
251 | delta = 1;
|
---|
252 | operation = 0;
|
---|
253 | count = 0;
|
---|
254 | while (!balance && count < 200) {
|
---|
255 | ilow = 0;
|
---|
256 | ihigh = 0;
|
---|
257 | count++;
|
---|
258 | for (j = 0; j < m; j++) {
|
---|
259 | if (distances[i, j] < mediana[i]) ilow++;
|
---|
260 | else if (distances[i, j] > mediana[i]) ihigh++;
|
---|
261 | }
|
---|
262 | if (ilow > ((m + 1) / 2)) {
|
---|
263 | mediana[i] = mediana[i] - delta;
|
---|
264 | if (operation == 1) delta = delta / 2;
|
---|
265 | operation = -1;
|
---|
266 | } else if (ihigh > ((m + 1) / 2)) {
|
---|
267 | mediana[i] = mediana[i] + delta;
|
---|
268 | if (operation == -1) delta = delta / 2;
|
---|
269 | operation = 1;
|
---|
270 | } else balance = true;
|
---|
271 | }
|
---|
272 | }
|
---|
273 | return mediana;
|
---|
274 | }
|
---|
275 | }
|
---|
276 | }
|
---|