org.jscience.computing.ai.evolutionary.geneticalgorithms
Class BinaryCodedGA

java.lang.Object
  org.jscience.computing.ai.evolutionary.geneticalgorithms.BinaryCodedGA

public abstract class BinaryCodedGA
extends java.lang.Object

Implementation of binary-coded genetic algorithm.

If you don't want to use default implementation of any method. You should extend this class and override that method.

Field Summary

protected double	avgFitness Average fitness in the current population
protected int	bestChromosomeNo Index of the best chromosome in the current generation
protected int	chromosomeLength Size of the chromosomes for individuals
protected double	crossoverProbability Probability of crossover operation
protected boolean[][]	currentPopulation Individuals in the current generation
protected double[]	currentPopulationFitness Fitnesses of the individuals in the current generation
protected int	generationNo Current generation number
protected double[]	historyAvgFitness Average fitnesses over all generations
protected int	historyLength Maxiumum hist�ry length for historyMaxFitness, historyMinFitness and historyAvgFitness variables
protected double[]	historyMaxFitness Maximum fitnesses over all generations
protected double[]	historyMinFitness Minimum fitnesses over all generations
protected double	maxFitness Maximum fitness in the current population
protected double	minFitness Minimum fitness in the current population
protected double	mutationProbability Probability of mutation operation
protected boolean[][]	nextPopulation Individuals in the next generation
protected int	populationSize Size of the population
protected double	tmpSumOfFitnesses TODO: Remove this vriable in the future relases.
protected double	tmpSumx TODO: Remove this vriable in the future relases.

Constructor Summary

BinaryCodedGA()
Creates a binary-coded genetic algorithm with default parameters.

BinaryCodedGA(int populationSize, int chromosomeLength, double crossoverProbability, double mutationProbability)
Creates a binary-coded genetic algorithm with given parameters.

BinaryCodedGA(int populationSize, int chromosomeLength, double crossoverProbability, double mutationProbability, int maxHistory)
Creates a binary-coded genetic algorithm with given parameters.

Method Summary

protected void	crossoverChromosomes(boolean[] chromosome1, boolean[] chromosome2) Applies crossover operator to given chromosomes.
static boolean[]	deSerializeIndividual(java.lang.String fileName, boolean isBinary) Deserializes given individual from disk with the given file name.
double	evaluateIndividual(int chromosomeNo, boolean[] chromosome) Evaluates an individual in the population.
void	evolve() Evolves the population.
double	getAvgFitness() Returns average fitness in the current generation
double[]	getAvgHistory() Returns the history of the average fitnesses
int	getChromosomeLength() Returns the length of chromosomes.
double	getCrossoverProbability() Returns probability of the crossover operation
double[]	getCurrentPopulationFitness() Returns fitnesses of the individuals in the current generation
boolean[]	getFittestChromosome() Returns the best individual in the current generation
double	getMaxFitness() Returns maximum fitness in the current generation
double[]	getMaxHistory() Returns the fitness history of best individuals over all generations.
double	getMinFitness() Returns minimum fitness in the current generation
double[]	getMinHistory() Returns the fitness history of worst individuals over all generations.
double	getMutationProbability() Returns probability of the mutation operation
boolean[][]	getPopulation() Returns individuals of the current generation
int	getPopulationSize() Returns population size
void	initializePopulation() Intitializes the population.
protected boolean	mutateChromosome(boolean bit) Mutates a given bit in a chromosome randomly.
void	printIndividual(boolean[] chromosome) Prints given individual
protected int	selectChromosome() Selects a chromosome using roulette-selection.
static void	serializeIndividual(boolean[] chromosome, java.lang.String fileName, boolean isBinary) Serializes given individual to disk with the given file name.
void	setChromosomeFitness(int chromosomeNo, double newFitness) Changes fitness of the given chromosome
void	setCrossoverProbability(double newProbability) Changes probability of the crossover operation
void	setMutationProbability(double newProbability) Changes probability of the mutation operation

Methods inherited from class java.lang.Object

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

Field Detail

populationSize

protected int populationSize

Size of the population

chromosomeLength

protected int chromosomeLength

Size of the chromosomes for individuals

crossoverProbability

protected double crossoverProbability

Probability of crossover operation

mutationProbability

protected double mutationProbability

Probability of mutation operation

currentPopulation

protected boolean[][] currentPopulation

Individuals in the current generation

nextPopulation

protected boolean[][] nextPopulation

Individuals in the next generation

currentPopulationFitness

protected double[] currentPopulationFitness

Fitnesses of the individuals in the current generation

generationNo

protected int generationNo

Current generation number

bestChromosomeNo

protected int bestChromosomeNo

Index of the best chromosome in the current generation

maxFitness

protected double maxFitness

Maximum fitness in the current population

minFitness

protected double minFitness

Minimum fitness in the current population

avgFitness

protected double avgFitness

Average fitness in the current population

historyMaxFitness

protected double[] historyMaxFitness

Maximum fitnesses over all generations

historyMinFitness

protected double[] historyMinFitness

Minimum fitnesses over all generations

historyAvgFitness

protected double[] historyAvgFitness

Average fitnesses over all generations

historyLength

protected int historyLength

Maxiumum hist�ry length for historyMaxFitness, historyMinFitness and historyAvgFitness variables

tmpSumx

protected double tmpSumx

TODO: Remove this vriable in the future relases.

tmpSumOfFitnesses

protected double tmpSumOfFitnesses

TODO: Remove this vriable in the future relases.

Constructor Detail

BinaryCodedGA

public BinaryCodedGA()

Creates a binary-coded genetic algorithm with default parameters.

BinaryCodedGA

public BinaryCodedGA(int populationSize,
                     int chromosomeLength,
                     double crossoverProbability,
                     double mutationProbability)

Creates a binary-coded genetic algorithm with given parameters.

Parameters:

populationSize - size of the population

chromosomeLength - size of the chromosomes for individuals

crossoverProbability - probability of crossover operation

mutationProbability - probability of mutation operation

BinaryCodedGA

public BinaryCodedGA(int populationSize,
                     int chromosomeLength,
                     double crossoverProbability,
                     double mutationProbability,
                     int maxHistory)

Creates a binary-coded genetic algorithm with given parameters.

Parameters:

populationSize - size of the population

chromosomeLength - size of the chromosomes for individuals

crossoverProbability - probability of crossover operation

mutationProbability - probability of mutation operation

maxHistory - maxiumum hist�ry length for historyMaxFitness, historyMinFitness and historyAvgFitness variables

Method Detail

initializePopulation

public void initializePopulation()
                          throws InvalidFitnessValueException

Intitializes the population.

Individuals in the population are created randomly and their fitnesses are calculated.

If histories (historyAvgFitness ...) are used they are also updated.

Throws:

InvalidFitnessValueException - if the fitness value of any individual is lower than zero.

evaluateIndividual

public double evaluateIndividual(int chromosomeNo,
                                 boolean[] chromosome)

Evaluates an individual in the population.

Choromosome index parameter is used to update fitness related data of the individual. Ex. currentPopulationFitness

Parameters:

chromosomeNo - index of the chromosome in the current population to be evaluated.

chromosome - chromosome of the individual to be evaluated.

Returns:

fitness of the individual.

evolve

public void evolve()
            throws InvalidFitnessValueException

Evolves the population.

Throws:

InvalidFitnessValueException - if the fitness value of any individual is lower than zero.

selectChromosome

protected int selectChromosome()

Selects a chromosome using roulette-selection.

Returns:

index of the selected chromosome in the current population.

mutateChromosome

protected boolean mutateChromosome(boolean bit)

Mutates a given bit in a chromosome randomly.

If a randomly chosen number between zero and one is lower than mutationProbability then the bit is mutated.

Parameters:

bit - bit to mutate

Returns:

mutated bit

See Also:

mutationProbability

crossoverChromosomes

protected void crossoverChromosomes(boolean[] chromosome1,
                                    boolean[] chromosome2)

Applies crossover operator to given chromosomes.

Parameters:

chromosome1 - first parent chromosome

chromosome2 - second parent chromosome

setChromosomeFitness

public void setChromosomeFitness(int chromosomeNo,
                                 double newFitness)
                          throws InvalidFitnessValueException

Changes fitness of the given chromosome

Parameters:

chromosomeNo - index of the chromosome in the current population

newFitness - new fitness of the individual

Throws:

InvalidFitnessValueException - if a fitness value lower than zero is tried to be used.

getMaxFitness

public double getMaxFitness()

Returns maximum fitness in the current generation

Returns:

maximum fitness in the current generation

getMinFitness

public double getMinFitness()

Returns minimum fitness in the current generation

Returns:

minimum fitness in the current generation.

getAvgFitness

public double getAvgFitness()

Returns average fitness in the current generation

Returns:

average fitness in the current generation.

getCurrentPopulationFitness

public double[] getCurrentPopulationFitness()

Returns fitnesses of the individuals in the current generation

Returns:

fitnesses of the current population

getPopulation

public boolean[][] getPopulation()

Returns individuals of the current generation

Returns:

current population

getFittestChromosome

public boolean[] getFittestChromosome()

Returns the best individual in the current generation

Returns:

best individual

getMaxHistory

public double[] getMaxHistory()

Returns the fitness history of best individuals over all generations.

Returns:

fitnesses of the best individuals for all previous generations

getMinHistory

public double[] getMinHistory()

Returns the fitness history of worst individuals over all generations.

Returns:

fitnesses of the worst individuals for all previous generations

getAvgHistory

public double[] getAvgHistory()

Returns the history of the average fitnesses

Returns:

fitnesses of the average fitnesses for all previous generations

getCrossoverProbability

public double getCrossoverProbability()

Returns probability of the crossover operation

Returns:

probability of crossover

getMutationProbability

public double getMutationProbability()

Returns probability of the mutation operation

Returns:

probability of mutation

setCrossoverProbability

public void setCrossoverProbability(double newProbability)

Changes probability of the crossover operation

Parameters:

newProbability - new crossover probability

setMutationProbability

public void setMutationProbability(double newProbability)

Changes probability of the mutation operation

Parameters:

newProbability - new mutation probability

getPopulationSize

public int getPopulationSize()

Returns population size

Returns:

population size

getChromosomeLength

public int getChromosomeLength()

Returns the length of chromosomes.

Returns:

number of bits in the chromosome

printIndividual

public void printIndividual(boolean[] chromosome)

Prints given individual

Parameters:

chromosome - individual to print

serializeIndividual

public static void serializeIndividual(boolean[] chromosome,
                                       java.lang.String fileName,
                                       boolean isBinary)
                                throws java.io.IOException

Serializes given individual to disk with the given file name.

Parameters:

chromosome - individual to serialize

fileName - name of the file where the serialized individual will be written

isBinary - specifies whether the serialization done with Java's own serialization method or with a simple text representation. If set to true, then Java serialization will be used.

Throws:

java.io.IOException - if serialization fails

deSerializeIndividual

public static boolean[] deSerializeIndividual(java.lang.String fileName,
                                              boolean isBinary)
                                       throws java.io.IOException,
                                              java.lang.ClassNotFoundException

Deserializes given individual from disk with the given file name.

Parameters:

fileName - name of the file where the serialized individual will be written

isBinary - specifies whether the serialization done with Java's own serialization method or with a simple text representation. If set to true, then Java serialization will be used.

Returns:

deserialized individual

Throws:

java.lang.ClassNotFoundException - if deserialization fails

java.io.IOException - if deserialization fails