org.uncommons.watchmaker.framework.operators
Class AbstractCrossover<T>

java.lang.Object
  org.uncommons.watchmaker.framework.operators.AbstractCrossover<T>

Type Parameters:

T - The type of evolved candidates that are operated on by this cross-over implementation.

All Implemented Interfaces:

EvolutionaryOperator<T>

Direct Known Subclasses:

BitStringCrossover, ByteArrayCrossover, CharArrayCrossover, DoubleArrayCrossover, IntArrayCrossover, ListCrossover, ListOrderCrossover, ObjectArrayCrossover, StringCrossover

public abstract class AbstractCrossover<T>
extends Object
implements EvolutionaryOperator<T>

Generic base class for cross-over implementations. Supports all cross-over processes that operate on a pair of parent candidates.

Author:

Daniel Dyer

Constructor Summary

protected	AbstractCrossover(int crossoverPoints) Sets up a fixed-point cross-over implementation.
protected	AbstractCrossover(int crossoverPoints, Probability crossoverProbability) Sets up a cross-over implementation that uses a fixed number of cross-over points.
protected	AbstractCrossover(NumberGenerator<Integer> crossoverPointsVariable) Sets up a cross-over implementation that uses a variable number of cross-over points.
protected	AbstractCrossover(NumberGenerator<Integer> crossoverPointsVariable, NumberGenerator<Probability> crossoverProbabilityVariable) Sets up a cross-over implementation that uses a variable number of cross-over points.

Method Summary

List<T>	apply(List<T> selectedCandidates, Random rng) Applies the cross-over operation to the selected candidates.
protected abstract List<T>	mate(T parent1, T parent2, int numberOfCrossoverPoints, Random rng) Perform cross-over on a pair of parents to generate a pair of offspring.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

AbstractCrossover

protected AbstractCrossover(int crossoverPoints)

Sets up a fixed-point cross-over implementation. Cross-over is applied to all pairs of parents. To apply cross-over only to a proportion of parent pairs, use the AbstractCrossover(int, Probability) constructor.

Parameters:

crossoverPoints - The constant number of cross-over points to use for all cross-over operations.

AbstractCrossover

protected AbstractCrossover(int crossoverPoints,
                            Probability crossoverProbability)

Sets up a cross-over implementation that uses a fixed number of cross-over points. Cross-over is applied to a proportion of selected parent pairs, with the remainder copied unchanged into the output population. The size of this evolved proportion is controlled by the crossoverProbability parameter.

Parameters:

crossoverPoints - The constant number of cross-over points to use for all cross-over operations.

crossoverProbability - The probability that, once selected, a pair of parents will be subjected to cross-over rather than being copied, unchanged, into the output population. Must be in the range 0 < crossoverProbability <= 1

AbstractCrossover

protected AbstractCrossover(NumberGenerator<Integer> crossoverPointsVariable)

Sets up a cross-over implementation that uses a variable number of cross-over points. Cross-over is applied to all pairs of parents. To apply cross-over only to a proportion of parent pairs, use the AbstractCrossover(NumberGenerator, NumberGenerator) constructor.

Parameters:

crossoverPointsVariable - A random variable that provides a number of cross-over points for each cross-over operation.

AbstractCrossover

protected AbstractCrossover(NumberGenerator<Integer> crossoverPointsVariable,
                            NumberGenerator<Probability> crossoverProbabilityVariable)

Sets up a cross-over implementation that uses a variable number of cross-over points. Cross-over is applied to a proportion of selected parent pairs, with the remainder copied unchanged into the output population. The size of this evolved proportion is controlled by the crossoverProbabilityVariable parameter.

Parameters:

crossoverPointsVariable - A variable that provides a (possibly constant, possibly random) number of cross-over points for each cross-over operation.

crossoverProbabilityVariable - A variable that controls the probability that, once selected, a pair of parents will be subjected to cross-over rather than being copied, unchanged, into the output population.

Method Detail

apply

public List<T> apply(List<T> selectedCandidates,
                     Random rng)

Applies the cross-over operation to the selected candidates. Pairs of candidates are chosen randomly and subjected to cross-over to produce a pair of offspring candidates.

Specified by:

apply in interface EvolutionaryOperator<T>

Parameters:

selectedCandidates - The evolved individuals that have survived to be eligible to reproduce.

rng - A source of randomness used to determine the location of cross-over points.

Returns:

The combined set of evolved offspring generated by applying cross-over to the the selected candidates.

mate

protected abstract List<T> mate(T parent1,
                                T parent2,
                                int numberOfCrossoverPoints,
                                Random rng)

Perform cross-over on a pair of parents to generate a pair of offspring.

Parameters:

parent1 - One of two individuals that provides the source material for generating offspring.

parent2 - One of two individuals that provides the source material for generating offspring.

numberOfCrossoverPoints - The number of cross-overs performed on the two parents.

rng - A source of randomness used to determine the location of cross-over points.

Returns:

A list containing two evolved offspring.