com.softtechdesign.ga
Class GAStringsSeq

java.lang.Object
  |
  +--com.softtechdesign.ga.GA
        |
        +--com.softtechdesign.ga.GAStringsSeq

All Implemented Interfaces:

java.lang.Runnable

public abstract class GAStringsSeq

extends GA

  GAStringsSeq models chromosomes as arrays of strings.

  Chromosome strings store chromosomeDim genes. Unlike GAString (where each gene is a
  single char), with GAStringsSeq each gene itself is a string. For example, one might look
  like:
  "dw|l80|nop|dn2" --4 genes, each separated by a pipe char (nop = special no operation gene)

  This type of GA could be used for:
    determining a sequence for navigating a maze
    discovering a useful image processing algorithm (seq) for highlighting an image feature
    modeling any system where genes are naturally modeled as Strings
 

Author:

Jeff Smith jeff@SoftTechDesign.com

Field Summary

protected int	numPossGeneValues the number of possible gene values
protected java.lang.String[]	possGeneValues the building material or pool "allowed" or "possible" gene values

Fields inherited from class com.softtechdesign.ga.GA

bestFitnessChromIndex, chromNextGen, chromosomeDim, chromosomes, computeStatistics, crossoverProb, crossoverType, genAvgDeviation, genAvgFitness, maxGenerations, maxPrelimGenerations, mutationProb, numPrelimRuns, populationDim, prelimChrom, randomSelectionChance, worstFitnessChromIndex

Constructor Summary

GAStringsSeq(int chromosomeDim, int populationDim, double crossoverProb, int randomSelectionChance, int maxGenerations, int numPrelimRuns, int maxPrelimGenerations, double mutationProb, int chromDecPts, java.lang.String[] possGeneValues, int crossoverType, boolean computeStatistics)
Initialize the GAStringSeq

Method Summary

protected void	doOnePtCrossover(Chromosome Chrom1, Chromosome Chrom2) Genetically recombine the given chromosomes using a one point crossover technique.
protected void	doRandomMutation(int iChromIndex) Randomly reassign a gene in the chromosome identified by the given index (iChromIndex) to one of the possible gene values
protected void	doTwoPtCrossover(Chromosome Chrom1, Chromosome Chrom2) Genetically recombine the given chromosomes using a two point crossover technique which combines two chromosomes at two random genes (loci), creating two new chromosomes.
protected void	doUniformCrossover(Chromosome Chrom1, Chromosome Chrom2) Genetically recombine the given chromosomes using a uniform crossover technique.
protected ChromStrings	getChromosome(int index) Returns the chromosome given by index casted as a ChromStrings object
protected java.lang.String	getRandomGeneFromPossGenes() Randomly pick and return a gene value
protected void	initPopulation() Create random chromosomes from the given gene space.
private void	setGeneValue(int iChromIndex, int geneIndex, java.lang.String gene) Sets the value of a gene for the given chromosome at the given geneIndex

Methods inherited from class com.softtechdesign.ga.GA

addChromosomesToLog, binaryStrToInt, computeFitnessRankings, copyNextGenToThisGen, doGeneticMating, evolve, getAvgDeviation, getAvgDeviationAmongChroms, getAvgFitness, getAvgFitness, getChromosomeDim, getComputeStatistics, getCrossoverProb, getCrossoverType, getFitness, getFitnessRank, getFittestChromosome, getFittestChromosomesFitness, getMaxGenerations, getMaxPrelimGenerations, getMutationProb, getNumPrelimRuns, getPopulationDim, getRandom, getRandom, getRandomSelectionChance, run, selectTwoParents

Methods inherited from class java.lang.Object

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

Field Detail

possGeneValues

protected java.lang.String[] possGeneValues

the building material or pool "allowed" or "possible" gene values

numPossGeneValues

protected int numPossGeneValues

the number of possible gene values

Constructor Detail

GAStringsSeq

public GAStringsSeq(int chromosomeDim,
                    int populationDim,
                    double crossoverProb,
                    int randomSelectionChance,
                    int maxGenerations,
                    int numPrelimRuns,
                    int maxPrelimGenerations,
                    double mutationProb,
                    int chromDecPts,
                    java.lang.String[] possGeneValues,
                    int crossoverType,
                    boolean computeStatistics)
             throws GAException

Initialize the GAStringSeq

Parameters:

chromosomeDim -

populationDim -

crossoverProb -

randomSelectionChance -

maxGenerations -

numPrelimRuns -

maxPrelimGenerations -

mutationProb -

chromDecPts -

possGeneValues -

crossoverType -

computeStatistics -

Throws:

GAException -

Method Detail

getChromosome

protected ChromStrings getChromosome(int index)

Returns the chromosome given by index casted as a ChromStrings object

Parameters:

index -

Returns:

getRandomGeneFromPossGenes

protected java.lang.String getRandomGeneFromPossGenes()

Randomly pick and return a gene value

Returns:

String

doRandomMutation

protected void doRandomMutation(int iChromIndex)

Randomly reassign a gene in the chromosome identified by the given index (iChromIndex) to one of the possible gene values

Overrides:

doRandomMutation in class GA

Parameters:

iChromIndex -

setGeneValue

private void setGeneValue(int iChromIndex,
                          int geneIndex,
                          java.lang.String gene)

Sets the value of a gene for the given chromosome at the given geneIndex

Parameters:

iChromIndex -

geneIndex -

gene -

initPopulation

protected void initPopulation()

Create random chromosomes from the given gene space.

Overrides:

initPopulation in class GA

doOnePtCrossover

protected void doOnePtCrossover(Chromosome Chrom1,
                                Chromosome Chrom2)

Genetically recombine the given chromosomes using a one point crossover technique.

 For example, if we have:
   chromosome A = { "x1", "x2", "x3", "x4" }
   chromosome B = { "y1", "y2", "y3", "y4" }
 and we randomly choose the crossover point of 1, the new genes will be:
   new chromosome A = { "y1", "x2", "x3", "x4" }
   new chromosome B = { "x1", "y2", "y3", "y4" }
 

Overrides:

doOnePtCrossover in class GA

Parameters:

Chrom1 -

Chrom2 -

doTwoPtCrossover

protected void doTwoPtCrossover(Chromosome Chrom1,
                                Chromosome Chrom2)

Genetically recombine the given chromosomes using a two point crossover technique which combines two chromosomes at two random genes (loci), creating two new chromosomes.

 For example, if we have:
   chromosome A = { "x1", "x2", "x3", "x4", "x5" }
   chromosome B = { "y1", "y2", "y3", "y4", "y5" }
 and we randomly choose the crossover points of 1 and 3, the new genes will be:
   new chromosome A = { "y1", "x2", "y3", "x4", "x5" }
   new chromosome B = { "x1", "y2", "x3", "y4", "y5" }
 

Overrides:

doTwoPtCrossover in class GA

Parameters:

Chrom1 -

Chrom2 -

doUniformCrossover

protected void doUniformCrossover(Chromosome Chrom1,
                                  Chromosome Chrom2)

Genetically recombine the given chromosomes using a uniform crossover technique. This technique randomly swaps genes from one chromosome to another.

 For example, if we have:
   chromosome A = { "x1", "x2", "x3", "x4", "x5" }
   chromosome B = { "y1", "y2", "y3", "y4", "y5" }
 our uniform (random) crossover might result in something like:
   chromosome A = { "y1", "x2", "x3", "x4", "x5" }
   chromosome B = { "x1", "y2", "y3", "y4", "y5" }
 if only the first gene in the chromosome was swapped.
 

Overrides:

doUniformCrossover in class GA

Parameters:

Chrom1 -

Chrom2 -