001 /*
002 * Java Genetic Algorithm Library (jenetics-4.0.0).
003 * Copyright (c) 2007-2017 Franz Wilhelmstötter
004 *
005 * Licensed under the Apache License, Version 2.0 (the "License");
006 * you may not use this file except in compliance with the License.
007 * You may obtain a copy of the License at
008 *
009 * http://www.apache.org/licenses/LICENSE-2.0
010 *
011 * Unless required by applicable law or agreed to in writing, software
012 * distributed under the License is distributed on an "AS IS" BASIS,
013 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
014 * See the License for the specific language governing permissions and
015 * limitations under the License.
016 *
017 * Author:
018 * Franz Wilhelmstötter (franz.wilhelmstoetter@gmail.com)
019 */
020 package io.jenetics;
021
022 import static java.lang.String.format;
023 import static io.jenetics.internal.math.comb.subset;
024 import static io.jenetics.internal.math.random.indexes;
025
026 import java.util.Random;
027 import java.util.function.IntFunction;
028
029 import io.jenetics.util.MSeq;
030 import io.jenetics.util.RandomRegistry;
031 import io.jenetics.util.Seq;
032
033 /**
034 * <p>
035 * An enhanced genetic algorithm (EGA) combine elements of existing solutions in
036 * order to create a new solution, with some of the properties of each parent.
037 * Recombination creates a new chromosome by combining parts of two (or more)
038 * parent chromosomes. This combination of chromosomes can be made by selecting
039 * one or more crossover points, splitting these chromosomes on the selected
040 * points, and merge those portions of different chromosomes to form new ones.
041 * </p>
042 * <p>
043 * The recombination probability <i>P(r)</i> determines the probability that a
044 * given individual (genotype, not gene) of a population is selected for
045 * recombination. The (<i>mean</i>) number of changed individuals depend on the
046 * concrete implementation and can be vary from
047 * <i>P(r)</i>·<i>N<sub>G</sub></i> to
048 * <i>P(r)</i>·<i>N<sub>G</sub></i>·<i>O<sub>R</sub></i>, where
049 * <i>O<sub>R</sub></i> is the order of the recombination, which is the number
050 * of individuals involved int the {@link #recombine} method.
051 * </p>
052 *
053 * @author <a href="mailto:franz.wilhelmstoetter@gmail.com">Franz Wilhelmstötter</a>
054 * @since 1.0
055 * @version 4.0
056 */
057 public abstract class Recombinator<
058 G extends Gene<?, G>,
059 C extends Comparable<? super C>
060 >
061 extends AbstractAlterer<G, C>
062 {
063
064 private final int _order;
065
066 /**
067 * Constructs an alterer with a given recombination probability.
068 *
069 * @param probability The recombination probability.
070 * @param order the number of individuals involved in the
071 * {@link #recombine(MSeq, int[], long)} step
072 * @throws IllegalArgumentException if the {@code probability} is not in the
073 * valid range of {@code [0, 1]} or the given {@code order} is
074 * smaller than two.
075 */
076 protected Recombinator(final double probability, final int order) {
077 super(probability);
078 if (order < 2) {
079 throw new IllegalArgumentException(format(
080 "Order must be greater than one, but was %d.", order
081 ));
082 }
083 _order = order;
084 }
085
086 /**
087 * Return the number of individuals involved in the
088 * {@link #recombine(MSeq, int[], long)} step.
089 *
090 * @return the number of individuals involved in the recombination step.
091 */
092 public int getOrder() {
093 return _order;
094 }
095
096 @Override
097 public final AltererResult<G, C> alter(
098 final Seq<Phenotype<G, C>> population,
099 final long generation
100 ) {
101 final AltererResult<G, C> result;
102 if (population.size() >= 2) {
103 final Random random = RandomRegistry.getRandom();
104 final int order = Math.min(_order, population.size());
105
106 // Selection of the individuals for recombination.
107 final IntFunction<int[]> individuals = i -> {
108 final int[] ind = subset(population.size(), order, random);
109 ind[0] = i;
110 return ind;
111 };
112
113 final MSeq<Phenotype<G, C>> pop = MSeq.of(population);
114 final int count = indexes(random, population.size(), _probability)
115 .mapToObj(individuals)
116 .mapToInt(i -> recombine(pop, i, generation))
117 .sum();
118
119 result = AltererResult.of(pop.toISeq(), count);
120 } else {
121 result = AltererResult.of(population.asISeq());
122 }
123
124 return result;
125 }
126
127 /**
128 * Recombination template method. This method is called 0 to n times. It is
129 * guaranteed that this method is only called by one thread.
130 *
131 * @param population the population to recombine
132 * @param individuals the array with the indexes of the individuals which
133 * are involved in the <i>recombination</i> step. The length of the
134 * array is {@link #getOrder()}. The first individual is the
135 * <i>primary</i> individual.
136 * @param generation the current generation.
137 * @return the number of genes that has been altered.
138 */
139 protected abstract int recombine(
140 final MSeq<Phenotype<G, C>> population,
141 final int[] individuals,
142 final long generation
143 );
144
145 }
|