org.evosuite.ga.metaheuristics.mosa

Class AbstractMOSA<T extends Chromosome>

java.lang.Object

org.evosuite.ga.metaheuristics.GeneticAlgorithm<T>

org.evosuite.ga.metaheuristics.mosa.AbstractMOSA<T>

Type Parameters:

T -

All Implemented Interfaces:

Serializable, SearchAlgorithm

Direct Known Subclasses:

MOSA

public abstract class AbstractMOSA<T extends Chromosome>
extends GeneticAlgorithm<T>

Abstract class for MOSA

Author:

Annibale Panichella, Fitsum M. Kifetew

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
protected Ranking<T>	ranking Selected ranking strategy
protected SelectionFunction<T>	selectionFunction Selection function to select parents
protected TestSuiteFitnessFunction	suiteFitness keep track of overall suite fitness and coverage

Fields inherited from class org.evosuite.ga.metaheuristics.GeneticAlgorithm

archive, bloatControl, chromosomeFactory, crossoverFunction, currentIteration, fitnessFunctions, listeners, localObjective, localSearchProbability, population, populationLimit, stoppingConditions

Constructor Summary

Constructors

Constructor and Description
AbstractMOSA(ChromosomeFactory<T> factory) Constructor

Method Summary

Modifier and Type	Method and Description
protected boolean	areEqual(T test1, T test2) This method verifies whether two TestCromosome contain the same test case.
protected List<T>	breedNextGeneration() This method is used to generate new individuals (offsprings) from the current population
protected void	calculateFitness() Calculate fitness for the whole population
protected abstract void	calculateFitness(T c) This method computes the fitness scores only for the current goals
void	clearCachedResults(T chromosome) This method clears the cached results for a specific chromosome (e.g., fitness function values computed in previous generations).
protected abstract List<T>	getArchive()
List<T>	getBestIndividuals() Return the individual(s) with the highest fitChromosomeess
abstract Set<FitnessFunction<T>>	getCoveredGoals()
protected abstract List<T>	getFinalTestSuite()
protected List<T>	getNonDominatedSolutions(List<T> solutions) This method extracts non-dominated solutions (tests) according to all covered goal (e.g., branches)
void	initializePopulation() Set up initial population
protected void	notifyEvaluation(Chromosome chromosome) Notify all search listeners of fitness evaluation
protected abstract double	numberOfCoveredTargets()
boolean	removeUnusedVariables(T chromosome) When a test case is changed via crossover and/or mutation, it can contains some primitive variables that are not used as input (or to store the output) of method calls.

Methods inherited from class org.evosuite.ga.metaheuristics.GeneticAlgorithm

addBloatControl, addFitnessFunction, addFitnessFunctions, addListener, addStoppingCondition, applyLocalSearch, calculateFitnessAndSortPopulation, clearPopulation, disableFirstSecondaryCriterion, elitism, enableFirstSecondaryCriterion, evolve, generateInitialPopulation, generateRandomPopulation, generateSolution, getAge, getBestIndividual, getBudgetString, getFitnessFunction, getFitnessFunctions, getNumberOfFitnessFunctions, getPopulation, getPopulationSize, getSelectionFunction, getStoppingConditions, isBetterOrEqual, isFinished, isNextPopulationFull, isTooLong, notifyIteration, notifyMutation, notifySearchFinished, notifySearchStarted, printBudget, randomism, removeListener, removeStoppingCondition, resetStoppingConditions, setArchive, setBloatControl, setChromosomeFactory, setCrossOverFunction, setPopulationLimit, setSelectionFunction, setStoppingCondition, setStoppingConditionLimit, shouldApplyLocalSearch, sortPopulation, starveByFitness, starveRandomly, starveToLimit, toString, updateBestIndividualFromArchive, updateFitnessFunctionsAndValues, updateProbability, updateSecondaryCriterion, writeIndividuals

Methods inherited from class java.lang.Object

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

Field Detail

suiteFitness

protected TestSuiteFitnessFunction suiteFitness

keep track of overall suite fitness and coverage

selectionFunction

protected SelectionFunction<T extends Chromosome> selectionFunction

Selection function to select parents

ranking

protected Ranking<T extends Chromosome> ranking

Selected ranking strategy

Constructor Detail

AbstractMOSA

public AbstractMOSA(ChromosomeFactory<T> factory)

Constructor

Parameters:

factory - a ChromosomeFactory object

Method Detail

breedNextGeneration

protected List<T> breedNextGeneration()

This method is used to generate new individuals (offsprings) from the current population

Returns:

offspring population

clearCachedResults

public void clearCachedResults(T chromosome)

This method clears the cached results for a specific chromosome (e.g., fitness function values computed in previous generations). Since a test case is changed via crossover and/or mutation, previous data must be recomputed.

Parameters:

chromosome - TestChromosome to clean

removeUnusedVariables

public boolean removeUnusedVariables(T chromosome)

When a test case is changed via crossover and/or mutation, it can contains some primitive variables that are not used as input (or to store the output) of method calls. Thus, this method removes all these "trash" statements.

Parameters:

chromosome -

Returns:

true or false depending on whether "unused variables" are removed

notifyEvaluation

protected void notifyEvaluation(Chromosome chromosome)

Notify all search listeners of fitness evaluation

Overrides:

notifyEvaluation in class GeneticAlgorithm<T extends Chromosome>

Parameters:

chromosome - a Chromosome object.

calculateFitness

protected void calculateFitness()

Calculate fitness for the whole population

getBestIndividuals

public List<T> getBestIndividuals()

Description copied from class: GeneticAlgorithm

Return the individual(s) with the highest fitChromosomeess

Overrides:

getBestIndividuals in class GeneticAlgorithm<T extends Chromosome>

Returns:

a list of Chromosome object(s).

calculateFitness

protected abstract void calculateFitness(T c)

This method computes the fitness scores only for the current goals

Parameters:

c - chromosome

getFinalTestSuite

protected abstract List<T> getFinalTestSuite()

getArchive

protected abstract List<T> getArchive()

getNonDominatedSolutions

protected List<T> getNonDominatedSolutions(List<T> solutions)

This method extracts non-dominated solutions (tests) according to all covered goal (e.g., branches)

Parameters:

solutionSet - set of test cases to analyze with the "dominance" relationship

Returns:

the non-dominated set of test cases

areEqual

protected boolean areEqual(T test1,
                           T test2)

This method verifies whether two TestCromosome contain the same test case. Here the equality is computed looking at the strings composing the tests. This method is strongly needed in breedNextGeneration().

Parameters:

test1 - first test

test2 - second test

Returns:

true if the test1 and test 2 (meant as String) are equal to each other; false otherwise.

initializePopulation

public void initializePopulation()

Set up initial population

Specified by:

initializePopulation in class GeneticAlgorithm<T extends Chromosome>

numberOfCoveredTargets

protected abstract double numberOfCoveredTargets()

getCoveredGoals

public abstract Set<FitnessFunction<T>> getCoveredGoals()