Package de.sfuhrm.genetic

Class GeneticAlgorithm<H extends AbstractHypothesis<H>>

java.lang.Object

de.sfuhrm.genetic.GeneticAlgorithm<H>

Type Parameters:

H - The hypothesis class to use.

public class GeneticAlgorithm<H extends AbstractHypothesis<H>>
extends Object

Generic genetic algorithm implementation.

Author:

Stephan Fuhrmann

Constructor Summary

Constructors

Constructor	Description
GeneticAlgorithm(double inCrossOverRate, double inMutationRate, int inGenerationSize)	Constructs a new genetic algorithm.

Method Summary

Modifier and Type	Method	Description
protected void	crossover(List<H> population, Collection<H> selectedSet)	Cross-overs a fraction of crossOverRate hypothesis relative to their fitness.
Optional<H>	findMaximum(Function<H,Boolean> loopCondition, Supplier<H> hypothesisSupplier)	Perform the genetic operation.
protected Optional<H>	max(Collection<H> in)	Find the maximum fitness element of the given collection.
protected void	mutate(List<H> selectedSet)	Mutates a fraction of mutationRate hypothesis.
protected H	probabilisticSelect(List<H> population, Collection<H> targetList, boolean addToTargetList)	Probabilistically selects one hypothesis relative to the selection probability of it.
protected void	select(List<H> population, Collection<H> selectedList)	Selects a fraction of 1-crossOverRate hypothesis relative to their fitness.

Methods inherited from class java.lang.Object

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

Constructor Detail

GeneticAlgorithm

public GeneticAlgorithm(double inCrossOverRate,
                        double inMutationRate,
                        int inGenerationSize)

Constructs a new genetic algorithm.

Parameters:

inCrossOverRate - the fraction at which the cross over operator is applied to the population, between 0 and 1.

inMutationRate - the fraction at which the mutation operator is applied to the population, between 0 and 1.

inGenerationSize - the number of individual hypothesis in the population for each generation, greater than 1.

Throws:

IllegalArgumentException - if the parameters are illegal.

Method Detail

select

protected void select(List<H> population,
                      Collection<H> selectedList)

Selects a fraction of 1-crossOverRate hypothesis relative to their fitness.

Parameters:

population - the population to select on.

selectedList - the target list to put selected elements to.

crossover

protected void crossover(List<H> population,
                         Collection<H> selectedSet)

Cross-overs a fraction of crossOverRate hypothesis relative to their fitness.

Parameters:

population - the population to select on.

selectedSet - the target set to put crossed over elements to.

mutate

protected void mutate(List<H> selectedSet)

Mutates a fraction of mutationRate hypothesis.

Parameters:

selectedSet - the population to mutate on.

probabilisticSelect

protected H probabilisticSelect(List<H> population,
                                Collection<H> targetList,
                                boolean addToTargetList)

Probabilistically selects one hypothesis relative to the selection probability of it.

Parameters:

population - the population to select from.

targetList - the target list to eventually add the element to.

addToTargetList - whether to add the element to the targetList.

Returns:

the selected element.

max

protected Optional<H> max(Collection<H> in)

Find the maximum fitness element of the given collection.

Parameters:

in - the population to find the maximum in.

Returns:

the maximum element, if any.

findMaximum

public Optional<H> findMaximum(Function<H,Boolean> loopCondition,
                               Supplier<H> hypothesisSupplier)

Perform the genetic operation.

Parameters:

loopCondition - the abort condition that stays true while the maximum is not yet reached. Gets presented the best hypothesis as input.

hypothesisSupplier - creation function for new hypothesis.

Returns:

the maximum element, if any.