1 | #region License Information
|
---|
2 | /* HeuristicLab
|
---|
3 | * Copyright (C) 2002-2015 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 HeuristicLab.Data;
|
---|
25 | using HeuristicLab.Encodings.PermutationEncoding;
|
---|
26 | using HeuristicLab.Parameters;
|
---|
27 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
28 | using System;
|
---|
29 | namespace HeuristicLab.Problems.PTSP {
|
---|
30 | /// <summary>
|
---|
31 | /// An operator to evaluate inversion moves (2-opt).
|
---|
32 | /// </summary>
|
---|
33 | [Item("PTSPAnalyticalInversionMovePathEvaluator", "Evaluates an inversion move (2-opt) of the PTSP in with the closed form expression")]
|
---|
34 | [StorableClass]
|
---|
35 | public class PTSPAnalyticalInversionMovePathEvaluator : PTSPPathMoveEvaluator, IPermutationInversionMoveOperator {
|
---|
36 |
|
---|
37 | private static DoubleArray probabilities;
|
---|
38 | private static DoubleMatrix A;
|
---|
39 | private static DoubleMatrix B;
|
---|
40 |
|
---|
41 | public override Type EvaluatorType {
|
---|
42 | get { return typeof(PTSPAnalyticalInversionMovePathEvaluator); }
|
---|
43 | }
|
---|
44 | public ILookupParameter<InversionMove> InversionMoveParameter {
|
---|
45 | get { return (ILookupParameter<InversionMove>)Parameters["InversionMove"]; }
|
---|
46 | }
|
---|
47 |
|
---|
48 | public IValueParameter<DoubleArray> ProbabilitiesParameter {
|
---|
49 | get { return (IValueParameter<DoubleArray>)Parameters["Probabilities"]; }
|
---|
50 | }
|
---|
51 |
|
---|
52 | public IValueParameter<DoubleMatrix> AParameter {
|
---|
53 | get { return (IValueParameter<DoubleMatrix>)Parameters["A"]; }
|
---|
54 | }
|
---|
55 |
|
---|
56 | public IValueParameter<DoubleMatrix> BParameter {
|
---|
57 | get { return (IValueParameter<DoubleMatrix>)Parameters["B"]; }
|
---|
58 | }
|
---|
59 |
|
---|
60 | [StorableConstructor]
|
---|
61 | protected PTSPAnalyticalInversionMovePathEvaluator(bool deserializing) : base(deserializing) { }
|
---|
62 | protected PTSPAnalyticalInversionMovePathEvaluator(PTSPAnalyticalInversionMovePathEvaluator original, Cloner cloner) : base(original, cloner) { }
|
---|
63 | public PTSPAnalyticalInversionMovePathEvaluator()
|
---|
64 | : base() {
|
---|
65 | Parameters.Add(new LookupParameter<InversionMove>("InversionMove", "The move to evaluate."));
|
---|
66 | Parameters.Add(new ValueParameter<ItemList<DoubleValue>>("Probabilities", "The probabilities of the current instance"));
|
---|
67 | Parameters.Add(new ValueParameter<DoubleMatrix>("A", "The matrix A for delta evaluation"));
|
---|
68 | Parameters.Add(new ValueParameter<DoubleMatrix>("B", "The matrix B for delta evaluation"));
|
---|
69 | }
|
---|
70 |
|
---|
71 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
72 | return new PTSPAnalyticalInversionMovePathEvaluator(this, cloner);
|
---|
73 | }
|
---|
74 |
|
---|
75 | public static double EvaluateByCoordinates(Permutation permutation, InversionMove move, DoubleMatrix coordinates, PTSPAnalyticalInversionMovePathEvaluator evaluator) {
|
---|
76 | int edge1source = permutation.GetCircular(move.Index1 - 1);
|
---|
77 | int edge1target = permutation[move.Index1];
|
---|
78 | int edge2source = permutation[move.Index2];
|
---|
79 | int edge2target = permutation.GetCircular(move.Index2 + 1);
|
---|
80 | if (move.Index2 - move.Index1 >= permutation.Length - 2) return 0;
|
---|
81 | double moveQuality = 0;
|
---|
82 | // remove two edges
|
---|
83 | moveQuality -= evaluator.CalculateDistance(coordinates[edge1source, 0], coordinates[edge1source, 1],
|
---|
84 | coordinates[edge1target, 0], coordinates[edge1target, 1]);
|
---|
85 | moveQuality -= evaluator.CalculateDistance(coordinates[edge2source, 0], coordinates[edge2source, 1],
|
---|
86 | coordinates[edge2target, 0], coordinates[edge2target, 1]);
|
---|
87 | // add two edges
|
---|
88 | moveQuality += evaluator.CalculateDistance(coordinates[edge1source, 0], coordinates[edge1source, 1],
|
---|
89 | coordinates[edge2source, 0], coordinates[edge2source, 1]);
|
---|
90 | moveQuality += evaluator.CalculateDistance(coordinates[edge1target, 0], coordinates[edge1target, 1],
|
---|
91 | coordinates[edge2target, 0], coordinates[edge2target, 1]);
|
---|
92 | return moveQuality;
|
---|
93 | }
|
---|
94 |
|
---|
95 | public static double EvaluateByDistanceMatrix(Permutation permutation, InversionMove move, DistanceMatrix distanceMatrix) {
|
---|
96 | int i = move.Index1;
|
---|
97 | int j = move.Index2;
|
---|
98 | return RecursiveExpectedCost(1, i, j) + RecursiveExpectedCost(2, i, j) - RecursiveExpectedCost(3, i, j);
|
---|
99 | }
|
---|
100 |
|
---|
101 | protected static double RecursiveExpectedCost(int s, int i, int j) {
|
---|
102 | switch (s) {
|
---|
103 | case 1:
|
---|
104 | if (j == i + 1) {
|
---|
105 | return (1/(1-probabilities[i+1]))*A[i,2]+(1 - probabilities[i])*(A[i+1,1]-A[i+1,probabilities.Length-1]);
|
---|
106 | } else if (i == j) {
|
---|
107 | return A[i,1];
|
---|
108 | } else {
|
---|
109 | // Equation 25
|
---|
110 | return ((1 - probabilities[i]) / (1 - probabilities[j])) * RecursiveExpectedCost(1, i + 1, j - 1) + (1 - probabilities[i]) * (1);
|
---|
111 | }
|
---|
112 |
|
---|
113 | case 2:
|
---|
114 | if (j == i + 1) {
|
---|
115 | return (1 - probabilities[i + 1]) * (B[i, 1] - B[i, probabilities.Length - 1]) + (1 / (1 - probabilities[i])) * (B[i + 1, 2]);
|
---|
116 | } else if (i == j) {
|
---|
117 | return B[i,1];
|
---|
118 | } else {
|
---|
119 | return 0;
|
---|
120 | // Equation 26
|
---|
121 | }
|
---|
122 |
|
---|
123 | case 3:
|
---|
124 | if (j == i + 1) {
|
---|
125 | return A[i, 2] + A[i + 1, 1] - A[i + 1, probabilities.Length - 1] + B[i, 1] - B[i, probabilities.Length - 1] + B[i + 1, 2];
|
---|
126 | } else if (i == j) {
|
---|
127 | return A[i,1]+B[i,1];
|
---|
128 | } else {
|
---|
129 | return 0;
|
---|
130 | // Equation 27
|
---|
131 | }
|
---|
132 | default:
|
---|
133 | return 0;
|
---|
134 | }
|
---|
135 | }
|
---|
136 |
|
---|
137 | private double Q(int i, int j, DoubleArray probabilities) {
|
---|
138 | double prod = 1;
|
---|
139 | for (int k = i; k <= j; k++) {
|
---|
140 | prod *= (1 - probabilities[k]);
|
---|
141 | }
|
---|
142 | return prod;
|
---|
143 | }
|
---|
144 |
|
---|
145 | private double Q_hat(int i, int j, DoubleArray probabilities) {
|
---|
146 | double prod = 1;
|
---|
147 | for (int k = i; k <= j; k++) {
|
---|
148 | prod *= (1-probabilities[k]);
|
---|
149 | }
|
---|
150 | return prod;
|
---|
151 | }
|
---|
152 |
|
---|
153 | protected override double EvaluateByCoordinates(Permutation permutation, DoubleMatrix coordinates) {
|
---|
154 | return EvaluateByCoordinates(permutation, InversionMoveParameter.ActualValue, coordinates, this);
|
---|
155 | }
|
---|
156 |
|
---|
157 | protected override double EvaluateByDistanceMatrix(Permutation permutation, DistanceMatrix distanceMatrix) {
|
---|
158 | A = AParameter.Value;
|
---|
159 | B = BParameter.Value;
|
---|
160 | probabilities = ProbabilitiesParameter.Value;
|
---|
161 | return EvaluateByDistanceMatrix(permutation, InversionMoveParameter.ActualValue, distanceMatrix);
|
---|
162 | }
|
---|
163 |
|
---|
164 | protected override double CalculateDistance(double x1, double y1, double x2, double y2) {
|
---|
165 | return Math.Round(Math.Sqrt((x1 - x2) * (x1 - x2) + (y1 - y2) * (y1 - y2)));
|
---|
166 | }
|
---|
167 | }
|
---|
168 | }
|
---|