org.uma.jmetal.algorithm.multiobjective.abyss

Class ABYSS

java.lang.Object

org.uma.jmetal.algorithm.impl.AbstractScatterSearch<DoubleSolution,List<DoubleSolution>>

org.uma.jmetal.algorithm.multiobjective.abyss.ABYSS

All Implemented Interfaces:

Serializable, Runnable, Algorithm<List<DoubleSolution>>

public class ABYSS
extends AbstractScatterSearch<DoubleSolution,List<DoubleSolution>>

This class implements the AbYSS algorithm, a multiobjective scatter search metaheuristics, which is described in: A.J. Nebro, F. Luna, E. Alba, B. Dorronsoro, J.J. Durillo, A. Beham "AbYSS: Adapting Scatter Search to Multiobjective Optimization." IEEE Transactions on Evolutionary Computation. Vol. 12, No. 4 (August 2008), pp. 439-457

Author:

Antonio J. Nebro , Cristobal Barba

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
protected Archive<DoubleSolution>	archive
protected int	archiveSize
protected CrossoverOperator<DoubleSolution>	crossover
protected Comparator<Solution>	crowdingDistanceComparator
protected DistanceToSolutionListAttribute	distanceToSolutionListAttribute
protected Comparator<Solution>	dominanceComparator
protected Comparator<Solution>	equalComparator
protected int	evaluations
protected Comparator<Solution>	fitnessComparator
protected int[][]	frequency
protected LocalSearchOperator<DoubleSolution>	localSearch
protected MarkAttribute	marked
protected int	maxEvaluations
protected int	numberOfSubRanges These variables are used in the diversification method.
protected Problem<DoubleSolution>	problem
protected JMetalRandom	randomGenerator
protected List<DoubleSolution>	referenceSet1
protected int	referenceSet1Size
protected List<DoubleSolution>	referenceSet2
protected int	referenceSet2Size
protected int[][]	reverseFrequency
protected StrengthRawFitness	strengthRawFitness
protected int[]	sumOfFrequencyValues
protected int[]	sumOfReverseFrequencyValues

Constructor Summary

Constructors

Constructor and Description
ABYSS(DoubleProblem problem, int maxEvaluations, int populationSize, int referenceSet1Size, int referenceSet2Size, int archiveSize, Archive<DoubleSolution> archive, LocalSearchOperator<DoubleSolution> localSearch, CrossoverOperator<DoubleSolution> crossoverOperator, int numberOfSubRanges)

Method Summary

Modifier and Type	Method and Description
void	buildNewReferenceSet1() Build the referenceSet1 by moving the best referenceSet1Size individuals, according to a fitness comparator, from the population to the referenceSet1
void	buildNewReferenceSet2() Build the referenceSet2 by moving to it the most diverse referenceSet2Size individuals from the population in respect to the referenceSet1.
DoubleSolution	diversificationGeneration()
List<List<DoubleSolution>>	generatePairsFromSolutionList(List<DoubleSolution> solutionList) Generate all pair combinations of the referenceSet1
List<DoubleSolution>	getResult()
DoubleSolution	improvement(DoubleSolution solution)
boolean	isStoppingConditionReached()
void	referenceSetUpdate() Build the reference set after the initialization phase
void	referenceSetUpdate(DoubleSolution solution) Update the reference set with a new solution
boolean	refSet1Test(DoubleSolution solution) Tries to update the reference set one with a solution
boolean	refSet2Test(DoubleSolution solution) Tries to update the reference set 2 with a Solution
void	restart()
boolean	restartConditionIsFulfilled(List<DoubleSolution> combinedSolutions)
List<DoubleSolution>	solutionCombination(List<List<DoubleSolution>> solutionList)
List<List<DoubleSolution>>	subsetGeneration()

Methods inherited from class org.uma.jmetal.algorithm.impl.AbstractScatterSearch

getPopulation, getPopulationSize, initializationPhase, run, setPopulation, setPopulationSize

Methods inherited from class java.lang.Object

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

Field Detail

maxEvaluations

protected final int maxEvaluations

problem

protected final Problem<DoubleSolution> problem

referenceSet1Size

protected final int referenceSet1Size

referenceSet2Size

protected final int referenceSet2Size

referenceSet1

protected List<DoubleSolution> referenceSet1

referenceSet2

protected List<DoubleSolution> referenceSet2

archiveSize

protected final int archiveSize

archive

protected Archive<DoubleSolution> archive

localSearch

protected LocalSearchOperator<DoubleSolution> localSearch

crossover

protected CrossoverOperator<DoubleSolution> crossover

evaluations

protected int evaluations

randomGenerator

protected JMetalRandom randomGenerator

numberOfSubRanges

protected int numberOfSubRanges

These variables are used in the diversification method.

sumOfFrequencyValues

protected int[] sumOfFrequencyValues

sumOfReverseFrequencyValues

protected int[] sumOfReverseFrequencyValues

frequency

protected int[][] frequency

reverseFrequency

protected int[][] reverseFrequency

strengthRawFitness

protected StrengthRawFitness strengthRawFitness

fitnessComparator

protected Comparator<Solution> fitnessComparator

marked

protected MarkAttribute marked

distanceToSolutionListAttribute

protected DistanceToSolutionListAttribute distanceToSolutionListAttribute

dominanceComparator

protected Comparator<Solution> dominanceComparator

equalComparator

protected Comparator<Solution> equalComparator

crowdingDistanceComparator

protected Comparator<Solution> crowdingDistanceComparator

Constructor Detail

ABYSS

public ABYSS(DoubleProblem problem,
             int maxEvaluations,
             int populationSize,
             int referenceSet1Size,
             int referenceSet2Size,
             int archiveSize,
             Archive<DoubleSolution> archive,
             LocalSearchOperator<DoubleSolution> localSearch,
             CrossoverOperator<DoubleSolution> crossoverOperator,
             int numberOfSubRanges)

Method Detail

isStoppingConditionReached

public boolean isStoppingConditionReached()

Specified by:

isStoppingConditionReached in class AbstractScatterSearch<DoubleSolution,List<DoubleSolution>>

improvement

public DoubleSolution improvement(DoubleSolution solution)

Specified by:

improvement in class AbstractScatterSearch<DoubleSolution,List<DoubleSolution>>

getResult

public List<DoubleSolution> getResult()

Specified by:

getResult in interface Algorithm<List<DoubleSolution>>

Specified by:

getResult in class AbstractScatterSearch<DoubleSolution,List<DoubleSolution>>

diversificationGeneration

public DoubleSolution diversificationGeneration()

Specified by:

diversificationGeneration in class AbstractScatterSearch<DoubleSolution,List<DoubleSolution>>

referenceSetUpdate

public void referenceSetUpdate()

Build the reference set after the initialization phase

Specified by:

referenceSetUpdate in class AbstractScatterSearch<DoubleSolution,List<DoubleSolution>>

referenceSetUpdate

public void referenceSetUpdate(DoubleSolution solution)

Update the reference set with a new solution

Specified by:

referenceSetUpdate in class AbstractScatterSearch<DoubleSolution,List<DoubleSolution>>

Parameters:

solution -

buildNewReferenceSet1

public void buildNewReferenceSet1()

Build the referenceSet1 by moving the best referenceSet1Size individuals, according to a fitness comparator, from the population to the referenceSet1

buildNewReferenceSet2

public void buildNewReferenceSet2()

Build the referenceSet2 by moving to it the most diverse referenceSet2Size individuals from the population in respect to the referenceSet1. The size of the referenceSet2 can be lower than referenceSet2Size depending on the current size of the population

refSet1Test

public boolean refSet1Test(DoubleSolution solution)

Tries to update the reference set one with a solution

Parameters:

solution - The Solution

Returns:

true if the Solution has been inserted, false otherwise.

refSet2Test

public boolean refSet2Test(DoubleSolution solution)

Tries to update the reference set 2 with a Solution

Parameters:

solution - The Solution

Returns:

true if the Solution has been inserted, false otherwise.

Throws:

JMException

restartConditionIsFulfilled

public boolean restartConditionIsFulfilled(List<DoubleSolution> combinedSolutions)

Specified by:

restartConditionIsFulfilled in class AbstractScatterSearch<DoubleSolution,List<DoubleSolution>>

subsetGeneration

public List<List<DoubleSolution>> subsetGeneration()

Specified by:

subsetGeneration in class AbstractScatterSearch<DoubleSolution,List<DoubleSolution>>

Returns:

generatePairsFromSolutionList

public List<List<DoubleSolution>> generatePairsFromSolutionList(List<DoubleSolution> solutionList)

Generate all pair combinations of the referenceSet1

solutionCombination

public List<DoubleSolution> solutionCombination(List<List<DoubleSolution>> solutionList)

Specified by:

solutionCombination in class AbstractScatterSearch<DoubleSolution,List<DoubleSolution>>

restart

public void restart()

Specified by:

restart in class AbstractScatterSearch<DoubleSolution,List<DoubleSolution>>