1 | #region License Information
|
---|
2 | /* HeuristicLab
|
---|
3 | * Copyright (C) 2002-2018 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 HeuristicLab.Common;
|
---|
26 |
|
---|
27 | namespace HeuristicLab.Analysis.FitnessLandscape {
|
---|
28 | /// <summary>
|
---|
29 | /// A generic implementation of a directed walk that can still be used in an efficient manner.
|
---|
30 | /// </summary>
|
---|
31 | /// <remarks>
|
---|
32 | /// The algorithm is only implemented for use in the API
|
---|
33 | /// </remarks>
|
---|
34 | /// <typeparam name="T">The type of the solution encoding.</typeparam>
|
---|
35 | public static class DirectedWalk<T> where T : IDeepCloneable {
|
---|
36 | /// <summary>
|
---|
37 | /// Provides regular directed walks between two solutions.
|
---|
38 | /// </summary>
|
---|
39 | /// <param name="maximization">Whether the problem is a maximization or minimization problem.</param>
|
---|
40 | /// <param name="distFunc">The distance function that compares two solutions.</param>
|
---|
41 | /// <param name="neighborFunc">The neighborhood functions that returns _all_ neighbors of a solution.</param>
|
---|
42 | /// <param name="start">The starting solution (to which distance should be increased in every step).</param>
|
---|
43 | /// <param name="startFitness">Fitness value of the starting solution.</param>
|
---|
44 | /// <param name="target">The destination solution (to which distance should be decreased in every step).</param>
|
---|
45 | /// <param name="firstImprovement">Whether the walk should choose the first improving neighbor instead of scanning the whole neighborhood.</param>
|
---|
46 | /// <returns>The trail of the directed walk including start and target solution.</returns>
|
---|
47 | public static IEnumerable<Tuple<T, double>> Walk(bool maximization, Func<T, T, double> distFunc, Func<Tuple<T, double>, IEnumerable<Tuple<T, double>>> neighborFunc, T start, double startFitness, T target, bool firstImprovement = false) {
|
---|
48 | Func<T, T, bool> restriction = (current, neighbor) => distFunc(neighbor, target) < distFunc(current, target);
|
---|
49 | if (firstImprovement)
|
---|
50 | return DoFirstImprovementWalk(maximization, neighborFunc, start, startFitness, restriction);
|
---|
51 | return DoBestImprovementWalk(maximization, neighborFunc, start, startFitness, restriction);
|
---|
52 | }
|
---|
53 |
|
---|
54 | /// <summary>
|
---|
55 | /// Provides regular directed walks between two solutions when already a restricted neighborhood is given.
|
---|
56 | /// </summary>
|
---|
57 | /// <param name="maximization">Whether the problem is a maximization or minimization problem.</param>
|
---|
58 | /// <param name="restrictedNeighborFunc">The neighborhood functions that returns only those neighbors that move closer to some target solution.</param>
|
---|
59 | /// <param name="start">The starting solution.</param>
|
---|
60 | /// <param name="startFitness">Fitness value of the starting solution.</param>
|
---|
61 | /// <param name="firstImprovement">Whether the walk should choose the first improving neighbor instead of scanning the whole neighborhood.</param>
|
---|
62 | /// <returns>The trail of the directed walk including start solution (and potentially target solution if it was part of the final restricted neighborhood).</returns>
|
---|
63 | public static IEnumerable<Tuple<T, double>> WalkRestricted(bool maximization, Func<Tuple<T, double>, IEnumerable<Tuple<T, double>>> restrictedNeighborFunc, T start, double startFitness, bool firstImprovement = false) {
|
---|
64 | if (firstImprovement)
|
---|
65 | return DoFirstImprovementWalk(maximization, restrictedNeighborFunc, start, startFitness, (_, __) => true);
|
---|
66 | return DoBestImprovementWalk(maximization, restrictedNeighborFunc, start, startFitness, (_, __) => true);
|
---|
67 | }
|
---|
68 |
|
---|
69 | /// <summary>
|
---|
70 | /// Provides inverse directed walks away from a starting solution.
|
---|
71 | /// </summary>
|
---|
72 | /// <param name="maximization">Whether the problem is a maximization or minimization problem.</param>
|
---|
73 | /// <param name="distFunc">The distance function that compares two solutions.</param>
|
---|
74 | /// <param name="neighborFunc">The neighborhood functions that returns _all_ neighbors of a solution.</param>
|
---|
75 | /// <param name="start">The starting solution (to which distance should be increased in every step).</param>
|
---|
76 | /// <param name="startFitness">Fitness value of the starting solution.</param>
|
---|
77 | /// <param name="firstImprovement">Whether the walk should choose the first improving neighbor instead of scanning the whole neighborhood.</param>
|
---|
78 | /// <returns>The trail of the directed walk including start solution.</returns>
|
---|
79 | public static IEnumerable<Tuple<T, double>> InverseWalk(bool maximization, Func<T, T, double> distFunc, Func<Tuple<T, double>, IEnumerable<Tuple<T, double>>> neighborFunc, T start, double startFitness, bool firstImprovement = false) {
|
---|
80 | Func<T, T, bool> inverseRestriction = (current, neighbor) => distFunc(neighbor, start) > distFunc(current, start);
|
---|
81 | if (firstImprovement)
|
---|
82 | return DoFirstImprovementWalk(maximization, neighborFunc, start, startFitness, inverseRestriction);
|
---|
83 | return DoBestImprovementWalk(maximization, neighborFunc, start, startFitness, inverseRestriction);
|
---|
84 | }
|
---|
85 |
|
---|
86 | /// <summary>
|
---|
87 | /// Provides inverse directed walks away from a starting solution when already a restricted neighborhood is given.
|
---|
88 | /// </summary>
|
---|
89 | /// <param name="maximization">Whether the problem is a maximization or minimization problem.</param>
|
---|
90 | /// <param name="restrictedNeighborFunc">The neighborhood functions that returns only those neighbors that move away from the start solution.</param>
|
---|
91 | /// <param name="start">The starting solution.</param>
|
---|
92 | /// <param name="startFitness">Fitness value of the starting solution.</param>
|
---|
93 | /// <param name="firstImprovement">Whether the walk should choose the first improving neighbor instead of scanning the whole neighborhood.</param>
|
---|
94 | /// <returns>The trail of the directed walk including start solution.</returns>
|
---|
95 | public static IEnumerable<Tuple<T, double>> InverseWalkRestricted(bool maximization, Func<Tuple<T, double>, IEnumerable<Tuple<T, double>>> restrictedNeighborFunc, T start, double startFitness, bool firstImprovement = false) {
|
---|
96 | if (firstImprovement)
|
---|
97 | return DoFirstImprovementWalk(maximization, restrictedNeighborFunc, start, startFitness, (_, __) => true);
|
---|
98 | return DoBestImprovementWalk(maximization, restrictedNeighborFunc, start, startFitness, (_, __) => true);
|
---|
99 | }
|
---|
100 |
|
---|
101 | private static IEnumerable<Tuple<T, double>> DoBestImprovementWalk(bool maximization, Func<Tuple<T, double>, IEnumerable<Tuple<T, double>>> neighborFunc, T start, double startFitness, Func<T, T, bool> restriction) {
|
---|
102 | var current = Tuple.Create(start, startFitness);
|
---|
103 |
|
---|
104 | while (true) {
|
---|
105 | yield return current;
|
---|
106 |
|
---|
107 | var iterator = neighborFunc(current).Where(n => restriction(current.Item1, n.Item1)).GetEnumerator();
|
---|
108 | if (!iterator.MoveNext()) yield break;
|
---|
109 |
|
---|
110 | current = Tuple.Create((T)iterator.Current.Item1.Clone(), iterator.Current.Item2);
|
---|
111 |
|
---|
112 | while (iterator.MoveNext()) {
|
---|
113 | var other = iterator.Current;
|
---|
114 | if (maximization && other.Item2 > current.Item2 || !maximization && other.Item2 < current.Item2) {
|
---|
115 | current = Tuple.Create((T)other.Item1.Clone(), other.Item2);
|
---|
116 | }
|
---|
117 | }
|
---|
118 | }
|
---|
119 | }
|
---|
120 |
|
---|
121 | private static IEnumerable<Tuple<T, double>> DoFirstImprovementWalk(bool maximization, Func<Tuple<T, double>, IEnumerable<Tuple<T, double>>> neighborFunc, T start, double startFitness, Func<T, T, bool> restriction) {
|
---|
122 | var current = Tuple.Create(start, startFitness);
|
---|
123 |
|
---|
124 | while (true) {
|
---|
125 | yield return current;
|
---|
126 |
|
---|
127 | var iterator = neighborFunc(current).Where(n => restriction(current.Item1, n.Item1)).GetEnumerator();
|
---|
128 | if (!iterator.MoveNext()) yield break;
|
---|
129 |
|
---|
130 | var prevFitness = current.Item2;
|
---|
131 |
|
---|
132 | current = Tuple.Create((T)iterator.Current.Item1.Clone(), iterator.Current.Item2);
|
---|
133 |
|
---|
134 | if (maximization && current.Item2 > prevFitness || !maximization && current.Item2 < prevFitness)
|
---|
135 | continue;
|
---|
136 |
|
---|
137 | while (iterator.MoveNext()) {
|
---|
138 | var other = iterator.Current;
|
---|
139 | if (maximization && other.Item2 > current.Item2 || !maximization && other.Item2 < current.Item2) {
|
---|
140 | current = Tuple.Create((T)other.Item1.Clone(), other.Item2);
|
---|
141 |
|
---|
142 | if (maximization && current.Item2 > prevFitness || !maximization && current.Item2 < prevFitness)
|
---|
143 | break;
|
---|
144 | }
|
---|
145 | }
|
---|
146 | }
|
---|
147 | }
|
---|
148 | }
|
---|
149 | }
|
---|