org.generation5.bio
Class GeneticAlgorithm

java.lang.Object
  org.generation5.bio.GeneticAlgorithm

All Implemented Interfaces:

Steppable

public class GeneticAlgorithm

extends java.lang.Object

implements Steppable

GeneticAlgorithm will evolve anything that implements the Evolvable interface. The genetic algorithm implements Steppable, with each step of the GA being implemented within doStep.

Currently, only elitism is supported.

See Also:

Steppable, Steppable.doStep(), Evolvable

Field Summary

protected double	averageFitness The average fitness for this generation.
protected java.util.LinkedList	averageFitnesses The average fitness linked list.
protected java.util.LinkedList	bestFitnesses The best fitness linked list.
protected int	currentIteration The current iteration of the genetic algorithm, incremented upon each call to doStep.
protected double	elitePercentage The percentage of elites to copy to the new population (defaults to 0.05, or 5%).
protected boolean	logData Determines whether average and best fitness data should be logged.
protected int	maximumDataPoints The maximum number of data points to store in the linked lists.
protected int	maximumIterations The maximum number of iterations (currently unused).
protected double	mutationRate The mutation rate (default to 0.05, or 5%).
protected Evolvable[][]	population The population to evolve.

Constructor Summary

GeneticAlgorithm()
Creates a new instance of GeneticAlgorithm

GeneticAlgorithm(Evolvable[] population)
Create a instance of the genetic algorithm with an initial population.

Method Summary

void	doStep() Steps through one iteration of the genetic algorithm.
protected void	flipBuffer() Flips the genetic algorithm buffer.
double	getAverageFitness() Return the average fitness of this generation.
java.util.LinkedList	getAverageFitnesses() Return the average fitness linked list.
Evolvable	getBest() Retrieves the best Evolvable in the population.
double	getBestFitness() Retrieves the best fitness level.
java.util.LinkedList	getBestFitnesses() Return the best fitness linked list.
int	getCurrentIteration() Return the current iteration of the genetic algorithm.
Evolvable[]	getCurrentPopulation() Return the current population (double-buffering taken into account).
int	getMaximumDataPoints() Returns the maximum number of data points the GA is logging.
double	getMutationRate() Retrieve the current mutation rate.
void	init() Initialize the genetic algorithm.
protected Evolvable[]	inPopulation() The current population
protected Evolvable[]	outPopulation() The new population.
void	reset() Resets the genetic algorithm.
void	setMaximumDataPoints(int mdp) Set the maximum number of data points that the GA should log.
void	setMutationRate(double mr) Set the mutation rate (should generally be between 0.0 and 0.2).
void	setPopulation(Evolvable[] thisPopulation) Set the population to an initial array of Evolvables.

Methods inherited from class java.lang.Object

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

Field Detail

population

protected Evolvable[][] population

The population to evolve. The population is double-buffered, which increases speed but does require some caution when retrieving values from the class. See doStep and getBest for more information.

See Also:

doStep(), getBest()

currentIteration

protected int currentIteration

The current iteration of the genetic algorithm, incremented upon each call to doStep.

maximumIterations

protected int maximumIterations

The maximum number of iterations (currently unused).

elitePercentage

protected double elitePercentage

The percentage of elites to copy to the new population (defaults to 0.05, or 5%).

mutationRate

protected double mutationRate

The mutation rate (default to 0.05, or 5%).

averageFitness

protected double averageFitness

The average fitness for this generation.

logData

protected boolean logData

Determines whether average and best fitness data should be logged.

maximumDataPoints

protected int maximumDataPoints

The maximum number of data points to store in the linked lists.

averageFitnesses

protected java.util.LinkedList averageFitnesses

The average fitness linked list. Fitnesses are stored as PlotPoints for easy plotting.

See Also:

PlotPoint, Plot2D

bestFitnesses

protected java.util.LinkedList bestFitnesses

The best fitness linked list. Fitnesses are stored as PlotPoints for easy plotting.

See Also:

PlotPoint, Plot2D

Constructor Detail

GeneticAlgorithm

public GeneticAlgorithm()

Creates a new instance of GeneticAlgorithm

GeneticAlgorithm

public GeneticAlgorithm(Evolvable[] population)

Create a instance of the genetic algorithm with an initial population.

Parameters:

population - the initial population.

Method Detail

getAverageFitnesses

public java.util.LinkedList getAverageFitnesses()

Return the average fitness linked list.

Returns:

the average fitness linked list.

getBestFitnesses

public java.util.LinkedList getBestFitnesses()

Return the best fitness linked list.

Returns:

the best fitness linked list.

setPopulation

public void setPopulation(Evolvable[] thisPopulation)

Set the population to an initial array of Evolvables.

Parameters:

thisPopulation - the population to assign to the genetic algorithm.

See Also:

Evolvable

getCurrentPopulation

public Evolvable[] getCurrentPopulation()

Return the current population (double-buffering taken into account).

Returns:

the current population.

flipBuffer

protected void flipBuffer()

Flips the genetic algorithm buffer.

inPopulation

protected Evolvable[] inPopulation()

The current population

Returns:

the current population.

outPopulation

protected Evolvable[] outPopulation()

The new population.

Returns:

the new population.

doStep

public void doStep()

Steps through one iteration of the genetic algorithm. Fitnesses are calculated for each member of the population, sorted by fitness and mated together. doStep also implements elitism and mutation.

Note that at the end of the iteration, flipBuffer is called. This means that the inPopulation returns the new, unsorted population and outPopulation returns the "current", sorted population.

Specified by:

doStep in interface Steppable

setMutationRate

public void setMutationRate(double mr)

Set the mutation rate (should generally be between 0.0 and 0.2).

Parameters:

mr - the mutation rate.

getMutationRate

public double getMutationRate()

Retrieve the current mutation rate.

Returns:

the mutation rate.

init

public void init()

Initialize the genetic algorithm.

Specified by:

init in interface Steppable

getBest

public Evolvable getBest()

Retrieves the best Evolvable in the population. It is assumed that getBest is called after doStep, therefore the best population member will be equal to outPopulation()[0].

Returns:

the best Evolvable in the population.

See Also:

outPopulation()

getBestFitness

public double getBestFitness()

Retrieves the best fitness level. This works the same as getBest, so the assumptions remain the same.

Returns:

the best fitness in the population.

getAverageFitness

public double getAverageFitness()

Return the average fitness of this generation.

Returns:

the average fitness.

reset

public void reset()

Resets the genetic algorithm. This just calls init.

Specified by:

reset in interface Steppable

See Also:

init()

getCurrentIteration

public int getCurrentIteration()

Return the current iteration of the genetic algorithm.

Returns:

the current iteration of the genetic algorithm.

setMaximumDataPoints

public void setMaximumDataPoints(int mdp)

Set the maximum number of data points that the GA should log.

Parameters:

mdp - maximum number of data points to log.

getMaximumDataPoints

public int getMaximumDataPoints()

Returns the maximum number of data points the GA is logging.

Returns:

the maximum data points.