lib.easyjava.ml.optimization.ga

Class GeneticAlgorithmOptimizer<T extends GeneticAlgorithmSolution>

java.lang.Object

lib.easyjava.ml.optimization.ga.GeneticAlgorithmOptimizer<T>

Type Parameters:

T - the type of solution to be optimized

public class GeneticAlgorithmOptimizer<T extends GeneticAlgorithmSolution>
extends java.lang.Object

This implements a Genetic Algorithm for optimization of arbitrary structures. It can run generation-by-generation optimization of any type implementing GeneticAlgorithmSolution correctly.

Author:

Rob Rua (robrua@alumni.cmu.edu)

Constructor Summary

Constructors

Constructor and Description
GeneticAlgorithmOptimizer(AbstractSolutionFactory<T> factory) Creates an optimizer which generates a randomized initial population with default population size, mutation, crossover, and elite rates.
GeneticAlgorithmOptimizer(int populationSize, AbstractSolutionFactory<T> factory) Creates an optimizer which generates a randomized initial population with default mutation, crossover, and elite rates.
GeneticAlgorithmOptimizer(int populationSize, double mutationRate, double crossoverRate, double elitePercentage, AbstractSolutionFactory<T> factory) Creates an optimizer which generates a randomized initial population.
GeneticAlgorithmOptimizer(java.util.List<T> population) Creates an optimizer which runs starting from the initially seeded population default mutation, crossover, and elite rates.
GeneticAlgorithmOptimizer(java.util.List<T> population, double mutationRate, double crossoverRate, double elitePercentage) Creates an optimizer which runs starting from the initially seeded population.

Method Summary

Modifier and Type	Method and Description
void	nextGeneration() Proceeds to the next generation
java.util.List<T>	population() Returns the current population ordered by fitness.
void	runFor(int generations) Proceeds the specified number of generations

Methods inherited from class java.lang.Object

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

Constructor Detail

GeneticAlgorithmOptimizer

public GeneticAlgorithmOptimizer(AbstractSolutionFactory<T> factory)

Creates an optimizer which generates a randomized initial population with default population size, mutation, crossover, and elite rates.

Parameters:

factory - the factory which produces random solutions for the initial population

GeneticAlgorithmOptimizer

public GeneticAlgorithmOptimizer(int populationSize,
                                 AbstractSolutionFactory<T> factory)

Creates an optimizer which generates a randomized initial population with default mutation, crossover, and elite rates.

Parameters:

populationSize - the number of solutions in each generation

factory - the factory which produces random solutions for the initial population

GeneticAlgorithmOptimizer

public GeneticAlgorithmOptimizer(int populationSize,
                                 double mutationRate,
                                 double crossoverRate,
                                 double elitePercentage,
                                 AbstractSolutionFactory<T> factory)

Creates an optimizer which generates a randomized initial population.

Parameters:

populationSize - the number of solutions in each generation

mutationRate - the chance for a new solution to randomly mutate

crossoverRate - the chance for a new solution to be generated by combination of two elite solutions, rather than cloned from one

elitePercentage - the percentage of solutions in each generation to be considered elite. The best elitePercentage solutions are elite.

factory - the factory which produces random solutions for the initial population

GeneticAlgorithmOptimizer

public GeneticAlgorithmOptimizer(java.util.List<T> population)

Creates an optimizer which runs starting from the initially seeded population default mutation, crossover, and elite rates.

Parameters:

population - the initial population to optimize

GeneticAlgorithmOptimizer

public GeneticAlgorithmOptimizer(java.util.List<T> population,
                                 double mutationRate,
                                 double crossoverRate,
                                 double elitePercentage)

Creates an optimizer which runs starting from the initially seeded population.

Parameters:

population - the initial population to optimize

mutationRate - the chance for a new solution to randomly mutate

crossoverRate - the chance for a new solution to be generated by combination of two elite solutions, rather than cloned from one

elitePercentage - the percentage of solutions in each generation to be considered elite. The best elitePercentage solutions are elite.

Method Detail

nextGeneration

public void nextGeneration()

Proceeds to the next generation

population

public java.util.List<T> population()

Returns the current population ordered by fitness.

Returns:

the current population

runFor

public void runFor(int generations)

Proceeds the specified number of generations

Parameters:

generations - the number of generations to move forward