Genetic Algorithm Common Interface. More...

Inheritance diagram for GeneticAlgo:

Public Member Functions
	GeneticAlgo ()
	Constructor.
virtual	~GeneticAlgo ()
	Destructor.
QVector< QVector< double > >	averageFits () const
	Return the Average objective values over generation done.
QVector< QVector< double > >	bestFits () const
	Return the Best objective values over generation done.
virtual QVector< Evaluation * >	evaluationPool ()=0
	Return the Evaluation Pool containing the Evaluations effectively used during evolution.
virtual Evaluation *	evaluationPrototype ()=0
	Returns the Evaluation object used as prototype to eventually generate other evaluators.
void	evolve (unsigned int generationToReach)
	Evolve Genome until reach the generation specified.
void	evolve ()
	Evolve Genome until condition isEvolutionEnded has been reached.
virtual void	finalize ()=0
	Finalize the Genetic Algorithm Process.
virtual void	gaStep ()=0
	Execute a single minor step and return.
unsigned int	generation () const
	Return the current generation number.
Genome *	genome ()
	Return the current Genome.
const Genome *	genome () const
	Return the current Genome.
virtual void	initialize ()=0
	Initialize the Genetic Algorithm Process.
bool	isEvaluationDone ()
	returns true if we have evaluated genomes for the current generation
bool	isEvolutionEnded ()
	returns true if the evolution process has reached the end
QVector< double >	meanHammingDist () const
	Return the Mean of Hamming distance over generation done.
int	numGenerations ()
	Return the number of Generations.
void	setGeneration (unsigned int gen)
	Set the generation number.
void	setGenome (Genome *gen)
	Set the Genome to use as current population.
void	setNumGenerations (int g)
	Set the number of Generations to do.
virtual void	skipEvaluation ()=0
	Skip the evaluation phase and put the algorithm in the state just after the evaluation.
QVector< double >	standardDeviationHammingDist () const
	Return the Standard Deviation of Hamming distance over generation done.
QVector< double >	varianceHammingDist () const
	Return the Variance of Hamming distance over generation done.
Protected Member Functions
void	updateStats ()
	Update statistical data of current generation and Genome (Mean, Variance, Standard Deviation)
Protected Attributes
QVector< QVector< double > >	avgfits
	Statistical Data: report the Average value for each objective for each generation.
QVector< QVector< double > >	bestfits
	Statistical Data: report the Best value for each objective for each generation.
bool	evaluationDone
	setted by implementors to signal that genomes have been evaluated when true means that the genomes of the current generation have been evaluated (we are at the end of the end of the generation, next step will be reproduction and start of a new generation), when false means that genomes haven't still not been evaluated (we are at the beginning of a generation, next step will be evaluation of genomes)
bool	evolutionEnd
	setted by implementors to signal the finish of evolution process when true means that evolution is ended
unsigned int	generationv
	current generation
Genome *	genomev
	current genome
QVector< double >	meanHdists
	Statistical Data: report the mean of all hamming distance between genotypes for each generation.
unsigned int	numGens
	number of generations
QVector< double >	stdHdists
	Statistical Data: report the standard variation of all hamming distance between genotypes for each generation.
QVector< double >	varHdists
	Statistical Data: report the variance of all hamming distance between genotypes for each generation.

Detailed Description

Genetic Algorithm Common Interface.

Description: Interface of the Genetic Algorithm Process
This class is highly abstract and there is no methods here for setting Selection, Crossover, Mutation and Reproduction processes in order to allow to each sub-class to choose which subset of operator to use.
Hence, it would be possible to create a GA which allow to specify only Mutation and fix the other in some pre-determined ways.

Warnings

Definition at line 46 of file geneticalgo.h.

Constructor & Destructor Documentation

GeneticAlgo ( )

Constructor.

Definition at line 27 of file geneticalgo.cpp.

~GeneticAlgo ( ) [virtual]

Destructor.

Definition at line 41 of file geneticalgo.cpp.

References GeneticAlgo::genomev.

Member Function Documentation

QVector< QVector< double > > averageFits ( ) const

Return the Average objective values over generation done.

Definition at line 104 of file geneticalgo.cpp.

References GeneticAlgo::avgfits.

QVector< QVector< double > > bestFits ( ) const

Return the Best objective values over generation done.

Warning:: each entry of the vector contains a vector of max values for each objective, but it does not means that exists a genotype scoring all values for all objectives

Definition at line 100 of file geneticalgo.cpp.

References GeneticAlgo::bestfits.

virtual QVector<Evaluation*> evaluationPool ( ) [pure virtual]

Return the Evaluation Pool containing the Evaluations effectively used during evolution.

Implemented in LaralGA, NSGA2, ParallelGA, SimpleGA, and StefanoSteadyStateGA.

virtual Evaluation* evaluationPrototype ( ) [pure virtual]

Returns the Evaluation object used as prototype to eventually generate other evaluators.

This object could or could not be used to compute the fitness of genotypes (depending on the specific genetic algorithm)

Implemented in LaralGA, NSGA2, ParallelGA, SimpleGA, and StefanoSteadyStateGA.

void evolve ( unsigned int generationToReach )

Evolve Genome until reach the generation specified.

This function exists after haveing evaluated individuals of the generationToReach generation

Warning:: this method is Blocking

Definition at line 45 of file geneticalgo.cpp.

References GeneticAlgo::evaluationDone, GeneticAlgo::gaStep(), GeneticAlgo::generationv, and GeneticAlgo::isEvolutionEnded().

void evolve ( )

Evolve Genome until condition isEvolutionEnded has been reached.

Warning:: this method is Blocking

Definition at line 52 of file geneticalgo.cpp.

References GeneticAlgo::gaStep(), and GeneticAlgo::isEvolutionEnded().

virtual void finalize ( ) [pure virtual]

Finalize the Genetic Algorithm Process.

In order to create specific Algorithms, re-implement this method

Implemented in LaralGA, NSGA2, ParallelGA, SimpleGA, and StefanoSteadyStateGA.

virtual void gaStep ( ) [pure virtual]

Execute a single minor step and return.

In order to create specific Algorithms, re-implement this method

Implemented in LaralGA, NSGA2, ParallelGA, SimpleGA, and StefanoSteadyStateGA.

Referenced by GeneticAlgo::evolve().

unsigned int generation ( ) const

Return the current generation number.

Definition at line 88 of file geneticalgo.cpp.

References GeneticAlgo::generationv.

Referenced by StefanoSteadyStateGA::gaStep(), SimpleGA::gaStep(), ParallelGA::gaStep(), NSGA2::gaStep(), LaralGA::gaStep(), Mutation::mutationRate(), Mutation::setMutationRate(), SimpleGA::skipEvaluation(), ParallelGA::skipEvaluation(), NSGA2::skipEvaluation(), and LaralGA::skipEvaluation().

Genome * genome ( )

Return the current Genome.

Definition at line 70 of file geneticalgo.cpp.

References GeneticAlgo::genomev.

Referenced by StefanoSteadyStateGA::configure(), SimpleGA::configure(), ParallelGA::configure(), NSGA2::configure(), LaralGA::configure(), StefanoSteadyStateGA::finalize(), StefanoSteadyStateGA::gaStep(), SimpleGA::gaStep(), ParallelGA::gaStep(), NSGA2::gaStep(), LaralGA::gaStep(), StefanoSteadyStateGA::initialize(), StefanoSteadyStateGA::save(), SimpleGA::save(), ParallelGA::save(), NSGA2::save(), and LaralGA::save().

const Genome * genome ( ) const

Return the current Genome.

Definition at line 74 of file geneticalgo.cpp.

References GeneticAlgo::genomev.

virtual void initialize ( ) [pure virtual]

Initialize the Genetic Algorithm Process.

In order to create specific Algorithms, re-implement this method

Implemented in LaralGA, NSGA2, ParallelGA, SimpleGA, and StefanoSteadyStateGA.

bool isEvaluationDone ( )

returns true if we have evaluated genomes for the current generation

Definition at line 62 of file geneticalgo.cpp.

References GeneticAlgo::evaluationDone.

bool isEvolutionEnded ( )

returns true if the evolution process has reached the end

Definition at line 58 of file geneticalgo.cpp.

References GeneticAlgo::evolutionEnd.

Referenced by GeneticAlgo::evolve().

QVector< double > meanHammingDist ( ) const

Return the Mean of Hamming distance over generation done.

Definition at line 108 of file geneticalgo.cpp.

References GeneticAlgo::meanHdists.

int numGenerations ( )

Return the number of Generations.

Definition at line 96 of file geneticalgo.cpp.

References GeneticAlgo::numGens.

Referenced by StefanoSteadyStateGA::configure(), SimpleGA::configure(), ParallelGA::configure(), NSGA2::configure(), LaralGA::configure(), StefanoSteadyStateGA::save(), SimpleGA::save(), ParallelGA::save(), NSGA2::save(), and LaralGA::save().

void setGeneration ( unsigned int gen )

Set the generation number.

Warning:: be careful on when you use this function. If you set the generation when isEvaluationDone() is true, the next generation to be evaluated will be gen + 1

Definition at line 78 of file geneticalgo.cpp.

References GeneticAlgo::generationv.

Referenced by StefanoSteadyStateGA::gaStep(), SimpleGA::gaStep(), ParallelGA::gaStep(), NSGA2::gaStep(), LaralGA::gaStep(), StefanoSteadyStateGA::initialize(), SimpleGA::initialize(), ParallelGA::initialize(), NSGA2::initialize(), and LaralGA::initialize().

void setGenome ( Genome * gen )

Set the Genome to use as current population.

Definition at line 66 of file geneticalgo.cpp.

References GeneticAlgo::genomev.

Referenced by StefanoSteadyStateGA::configure(), SimpleGA::configure(), ParallelGA::configure(), NSGA2::configure(), LaralGA::configure(), SimpleGA::gaStep(), ParallelGA::gaStep(), and NSGA2::gaStep().

void setNumGenerations ( int g )

Set the number of Generations to do.

Definition at line 92 of file geneticalgo.cpp.

References GeneticAlgo::numGens.

Referenced by StefanoSteadyStateGA::configure(), SimpleGA::configure(), ParallelGA::configure(), NSGA2::configure(), and LaralGA::configure().

virtual void skipEvaluation ( ) [pure virtual]

Skip the evaluation phase and put the algorithm in the state just after the evaluation.

This can be used after restarting an interrupted evolution to prevent re-evaluation of the genome that was loaded from file.

Implemented in LaralGA, NSGA2, ParallelGA, SimpleGA, and StefanoSteadyStateGA.

QVector< double > standardDeviationHammingDist ( ) const

Return the Standard Deviation of Hamming distance over generation done.

Definition at line 116 of file geneticalgo.cpp.

References GeneticAlgo::stdHdists.

void updateStats ( ) [protected]

Update statistical data of current generation and Genome (Mean, Variance, Standard Deviation)

This method should be called by the user when it re-implement GeneticAlgo just afert the evaluation of all Genotypes in the Genome

Warning:: remember to call it at the right moment, otherwise you will have wrong statistic data

Definition at line 120 of file geneticalgo.cpp.

References Genome::at(), GeneticAlgo::avgfits, GeneticAlgo::bestfits, GeneticAlgo::generationv, GeneticAlgo::genomev, GeneticAlgo::meanHdists, Genotype::numOfObjectives(), Genotype::objective(), Genome::size(), GeneticAlgo::stdHdists, and GeneticAlgo::varHdists.

Referenced by StefanoSteadyStateGA::gaStep(), SimpleGA::gaStep(), ParallelGA::gaStep(), NSGA2::gaStep(), and LaralGA::gaStep().

QVector< double > varianceHammingDist ( ) const

Return the Variance of Hamming distance over generation done.

Definition at line 112 of file geneticalgo.cpp.

References GeneticAlgo::varHdists.

Member Data Documentation

QVector< QVector<double> > avgfits [protected]

Statistical Data: report the Average value for each objective for each generation.

Definition at line 159 of file geneticalgo.h.

Referenced by GeneticAlgo::averageFits(), and GeneticAlgo::updateStats().

QVector< QVector<double> > bestfits [protected]

Statistical Data: report the Best value for each objective for each generation.

Definition at line 157 of file geneticalgo.h.

Referenced by GeneticAlgo::bestFits(), and GeneticAlgo::updateStats().

bool evaluationDone [protected]

setted by implementors to signal that genomes have been evaluated
when true means that the genomes of the current generation have been evaluated (we are at the end of the end of the generation, next step will be reproduction and start of a new generation), when false means that genomes haven't still not been evaluated (we are at the beginning of a generation, next step will be evaluation of genomes)

Definition at line 147 of file geneticalgo.h.

Referenced by GeneticAlgo::evolve(), StefanoSteadyStateGA::gaStep(), SimpleGA::gaStep(), ParallelGA::gaStep(), NSGA2::gaStep(), LaralGA::gaStep(), StefanoSteadyStateGA::initialize(), SimpleGA::initialize(), ParallelGA::initialize(), NSGA2::initialize(), LaralGA::initialize(), GeneticAlgo::isEvaluationDone(), StefanoSteadyStateGA::skipEvaluation(), SimpleGA::skipEvaluation(), ParallelGA::skipEvaluation(), NSGA2::skipEvaluation(), and LaralGA::skipEvaluation().