Instance Generation Algorithm. Known as Meta-Evolutionary Algorithm (EIG). This algorithm uses a Weighted Fitness Function (WFF) to evaluate how biased and diverse are the instances generated during the evolutionary process. The instances generated should be biased to the performance of the target algorithm in the portfolio but also should show some diversity between them. More...

#include <EIG.h>

Inheritance diagram for EIG< IP, IS, OP, OS >:

Public Types
using	Alg = AbstractEA< OS >

Public Member Functions
	EIG (const float &fitness=0.6f, const float &novelty=0.4f)
	Construct a new EIG object using a fitness ratio of 0.6 and novelty of 0.4 by default. Use the EIGBuilder instead of the constructor for safe experimentation. More...

void	run () override
	Interface to run the EIG for generating a set of instances for the optimisation problem defined (OP) More...

json	to_json () const override
	Creates the JSON representation of the EIG. More...

string	getName () const override
	Get the name of the algorithm. More...

string	getID () const override
	Get the ID. Similar to getName.

virtual Front< IS >	getResults () const override
	Returns a set of individuals as a result of the evolutionary method. The individual are instances for the OP. More...

const IP *	getInstanceProblem () const
	Get a raw pointer to the Instance Generation Problem. More...

void	setInstanceProblem (unique_ptr< IP > problem)
	Set the Instance Generation Problem. More...

vector< Alg * >	getPortfolio () const
	Get the portfolio of algorithms used in the evolutionary process. More...

void	setPortfolio (vector< unique_ptr< Alg >> &configs)
	Set the portfolio of algorithms to use in the evolutionary process. The order of the algorithms in the portfolio is crucial. Notice that the first algorithm in the portfolio will be considered the target and therefore the instances will be generated according to its performance. Also, it is important to know the order of the remaining algorithms in order to analysis the results obtained since not tags are included in the results at this moment. More...

float	getMutationRate () const
	Get the mutation rate used in EIG for mutation instances during the evolutionary process. More...

void	setMutationRate (float mRate)
	Set the mutation rate used in EIG for mutation instances during the evolutionary process. More...

float	getCrossoverRate () const
	Get the crossover rate used in EIG for applying the crossover operator to instances during the evolutionary process. More...

void	setCrossoverRate (float cRate)
	Set the crossover rate used in EIG for applying the crossover operator to instances during the evolutionary process. More...

int	getRepetitions () const
	Get the number of repetitions that each algorithm in the portfolio must perform over each instance at every generation of the EIG. More...

void	setRepetitions (int reps)
	Set the number of repetitions that each algorithm in the portfolio must perform over each instance at every generation of the EIG. Due to stochastic nature of most algorithms used in EIG the number of repetitions should vary between 10 to 30. For deterministic heuristics can be set to one. More...

Mutation< IS > *	getMutation () const
	Get the mutation operator used to mutate instances in EIG. More...

void	setMutation (unique_ptr< Mutation< IS >> mut)
	Set the mutation operator. Takes ownership of the pointer. More...

Crossover< IS > *	getCrossover () const
	Get the crossover operator used in EIG. More...

void	setCrossover (unique_ptr< Crossover< IS >> cx)
	Set the crossover operator. Takes ownership of the pointer. More...

Selection< IS > *	getSelection () const
	Get the selection operator. More...

void	setSelection (unique_ptr< Selection< IS >> sel)
	Set the selection operator. Takes ownership of the pointer. More...

void	setReplacement (unique_ptr< Replacement< IS >> rep)
	Set the selection operator. Takes ownership of the pointer. More...

InstanceFitness *	getEvaluation () const
	Get the evaluation approach for the instances. More...

void	setInstanceFitness (unique_ptr< InstanceFitness > eval)
	Set the evaluation approach for the instances. More...

int	getGenerations ()
	Get the number of generations that EIG must perform. More...

const NoveltySearch< IS > *	getNoveltySearch () const
	Get the Novelty Search approach used in EIG. More...

void	setNoveltySearch (unique_ptr< NoveltySearch< IS >> noveltySearch)
	Set the Novelty Search approach to use. More...

void	setWeightedFunction (const float &fW, const float &nW)
	Set the weights to use in the WFF. More...

Public Member Functions inherited from AbstractEA< IS >
	AbstractEA ()
	Construct a new Abstract EA object with default parameter values. More...

	AbstractEA (const int &maxEvals, const int &popsize)
	AbstractEA constructor which initialises the parameters maxEvalautions and populationSize. More...

	AbstractEA (unique_ptr< PopulationEvaluator< IS >>, const int &maxEvals, const int &popsize)
	AbstractEA constructor which initialises the parameters maxEvalautions, the populationSize and the evaluator approach. More...

int	getPopulationSize () const
	Get the population size. More...

void	setPopulationSize (int pSize)
	Setter to update the population size. More...

double	getElapsedTime () const
	Get the elapsed time of the evolutionary process. More...

const vector< IS > &	getPopulation () const
	Gets a reference of the population. More...

int	getMaxEvaluations () const
	Get the maximum number of evaluations to perform. More...

void	setMaxEvaluations (int maxEval)
	Set a new maximum number of evaluations to perform. More...

const Problem< IS > *	getProblem () const
	Gets a pointer to the problem which is being solved. More...

virtual void	setProblem (shared_ptr< Problem< IS >> prob)
	Set the new problem to solve. Uses a shared_ptr pointer which updates the reference counter. Does no take the ownership. More...

virtual void	setProblem (Problem< IS > *prob)
	Set the new problem to solve. Uses a raw pointer and takes the ownership of the object. More...

PopulationEvaluator< IS > *	getEvaluator () const
	Gets a pointer to the population evaluator system. More...

void	setEvaluator (unique_ptr< PopulationEvaluator< IS >> eval)
	Updates the population evaluator system with the unique_ptr. The method takes the ownership of the object behind. More...

int	getPerformedEvaluations () const
	Get the performed evaluations. More...

void	setPerformedEvaluations (int pEvals)
	Set the Performed Evaluations. Useful for the updateProgress method. More...

void	setPopulation (const vector< IS > &pop)
	Set the Population. More...

virtual Evolution< IS >	getEvolution () const
	Get the Evolution data. More...

int	getPrintingInterval () const
	Get the interval of checkpoints. More...

Protected Member Functions
void	initProgress () override
	Initialises the number of generations performed to 0. More...

void	updateProgress () override
	Updates the evolution by increasing the number of generations performed. More...

void	finishProgress () override
	Finishes the execution of the EIG. More...

bool	isStoppingConditionReached () override
	Checks whether the number of performed generations has reached the maximum allowed. More...

void	createInitialPopulation () override
	Creates a initial population of Problem Instances of the OP type. Uses the IP method "createSolutions". More...

void	evaluatePopulation (vector< IS > &individuals) override
	Evaluates the population of individuals with all the algorithms in the portfolio. More...

void	evaluateIndividual (IS &individual)
	Evaluates the given IS with all the algorithms in the porftolio. More...

virtual void	evaluationPhase (vector< IS > &individuals)
	Evaluation phase of the EIG All the stages necessary to evaluated the population of individuals are performed from this method. More...

virtual void	reproduction (IS &solution, IS &solution2)
	Executes the genetic operator over two instances to generate new individuals. More...

virtual void	replacement (vector< IS > &individuals)
	Replacement method for the population. In this case, the replacement is a simple generational approach, which also includes the comparison of individuals with the archive and the feature set of the NS. More...

virtual void	updateEvolution (vector< IS > &solutions)
	Updates the evolution results of the algorithm for the given checkpoint. More...

Protected Member Functions inherited from AbstractEA< IS >
virtual void	updateEvolution (vector< IS > &pop)

virtual void	updateEvolution (const int &checkpoint, vector< IS > &)

Protected Attributes
unique_ptr< IP >	instProb

vector< unique_ptr< Alg > >	algPortfolio

unique_ptr< Mutation< IS > >	mutation

unique_ptr< Crossover< IS > >	crossover

unique_ptr< Selection< IS > >	selection

unique_ptr< Replacement< IS > >	repOperator

float	mutationRate

float	crossoverRate

int	repetitions

unique_ptr< NoveltySearch< IS > >	noveltySearch

unique_ptr< InstanceFitness >	instanceFitness

unique_ptr< Weighted< IS > >	weightedFitness

AverageFitness< OS >	averageFitness

BestFitness< OS >	bestFitness

Protected Attributes inherited from AbstractEA< IS >
int	maxEvaluations

int	performedEvaluations

int	populationSize

vector< IS >	population

shared_ptr< Problem< IS > >	problem

unique_ptr< PopulationEvaluator< IS > >	evaluator

Evolution< IS >	evolution

AvgEvolution< IS >	avgEvolution

std::chrono::system_clock::time_point	startTime

std::chrono::system_clock::time_point	endTime

double	elapsedTime

int	nextCheckpoint

int	evolutionInterval

Additional Inherited Members
Static Public Attributes inherited from AbstractEA< IS >
static const int	DEFAULT_POPULATION_SIZE
	Default population size equal to 100 individuals. More...

static const int	DEFAULT_EVALUATIONS_LIMIT
	Default evaluation limit equal to 100000 evaluations. More...

static const std::string	NAME

static const std::string	MAX_EVALUATIONS

static const std::string	POP_SIZE

static const std::string	ELAPSED_TIME

static const std::string	EVALUATOR

static const int	EVOLUTION_SIZE

Detailed Description

template<typename IP, typename IS, typename OP, typename OS>
class EIG< IP, IS, OP, OS >

Instance Generation Algorithm. Known as Meta-Evolutionary Algorithm (EIG). This algorithm uses a Weighted Fitness Function (WFF) to evaluate how biased and diverse are the instances generated during the evolutionary process. The instances generated should be biased to the performance of the target algorithm in the portfolio but also should show some diversity between them.

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Member Typedef Documentation

◆ Alg

template<typename IP , typename IS , typename OP , typename OS >

using EIG< IP, IS, OP, OS >::Alg = AbstractEA<OS>

Acronym for an AbstractEA<OS>

Constructor & Destructor Documentation

◆ EIG()

template<typename IP , typename IS , typename OP , typename OS >

EIG< IP, IS, OP, OS >::EIG	(	const float &	fitness = `0.6f`,
		const float &	novelty = `0.4f`
	)

Construct a new EIG object using a fitness ratio of 0.6 and novelty of 0.4 by default. Use the EIGBuilder instead of the constructor for safe experimentation.

Creates a default EIG with only the evaluation as EasyInstances.

Parameters

fitness	Ratio of fitness for the WFF
novelty	Ratio of novelty for the WFF

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Member Function Documentation

◆ createInitialPopulation()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::createInitialPopulation

overrideprotectedvirtual

Creates a initial population of Problem Instances of the OP type. Uses the IP method "createSolutions".

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Implements AbstractEA< IS >.

Reimplemented in MapElites< IP, IS, OP, OS >.

◆ evaluateIndividual()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::evaluateIndividual ( IS & individual )

protected

Evaluates the given IS with all the algorithms in the porftolio.

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Parameters

individual IS individual to evaluate using all the configurations

◆ evaluatePopulation()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::evaluatePopulation ( vector< IS > & individuals )

overrideprotectedvirtual

Evaluates the population of individuals with all the algorithms in the portfolio.

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Parameters

individuals Population of IS to evaluate using the configurations

Implements AbstractEA< IS >.

Reimplemented in MapElites< IP, IS, OP, OS >.

◆ evaluationPhase()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::evaluationPhase ( vector< IS > & individuals )

protectedvirtual

Evaluation phase of the EIG All the stages necessary to evaluated the population of individuals are performed from this method.

First, each instance is solved by each algorithm in the portfolio (repeatdly if necessary) to construct the biasedFitness.
Second, we run the novelty search approach to evaluate the diversity of the solutions within the population.
The Weighted-Fitness-Function is used to calculate the final fitness for each individual.
Template Parameters

IP Type of Instance Problem to toSolve

IS Type of instances of the Optimisation Problem

OP Optimisation problem to toSolve with the configurations

OS Solutions to the OP type

Parameters

individuals

◆ finishProgress()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::finishProgress

overrideprotectedvirtual

Finishes the execution of the EIG.

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Implements AbstractEA< IS >.

◆ getCrossover()

template<typename IP , typename IS , typename OP , typename OS >

Crossover<IS>* EIG< IP, IS, OP, OS >::getCrossover ( ) const

inline

Get the crossover operator used in EIG.

Returns: a raw pointer to Crossover<IS>

◆ getCrossoverRate()

template<typename IP , typename IS , typename OP , typename OS >

float EIG< IP, IS, OP, OS >::getCrossoverRate ( ) const

inline

Get the crossover rate used in EIG for applying the crossover operator to instances during the evolutionary process.

Returns: float

◆ getEvaluation()

template<typename IP , typename IS , typename OP , typename OS >

InstanceFitness* EIG< IP, IS, OP, OS >::getEvaluation ( ) const

inline

Get the evaluation approach for the instances.

Returns: InstanceFitness*

◆ getGenerations()

template<typename IP , typename IS , typename OP , typename OS >

int EIG< IP, IS, OP, OS >::getGenerations ( )

inline

Get the number of generations that EIG must perform.

Returns

◆ getInstanceProblem()

template<typename IP , typename IS , typename OP , typename OS >

const IP* EIG< IP, IS, OP, OS >::getInstanceProblem ( ) const

inline

Get a raw pointer to the Instance Generation Problem.

Returns

◆ getMutation()

template<typename IP , typename IS , typename OP , typename OS >

Mutation<IS>* EIG< IP, IS, OP, OS >::getMutation ( ) const

inline

Get the mutation operator used to mutate instances in EIG.

Returns: a raw pointer to Mutation<IS>

◆ getMutationRate()

template<typename IP , typename IS , typename OP , typename OS >

float EIG< IP, IS, OP, OS >::getMutationRate ( ) const

inline

Get the mutation rate used in EIG for mutation instances during the evolutionary process.

Returns: float

◆ getName()

template<typename IP , typename IS , typename OP , typename OS >

string EIG< IP, IS, OP, OS >::getName ( ) const

inlineoverridevirtual

Get the name of the algorithm.

Returns: a string with the content EIG

Implements AbstractEA< IS >.

Reimplemented in MapElites< IP, IS, OP, OS >.

◆ getNoveltySearch()

template<typename IP , typename IS , typename OP , typename OS >

const NoveltySearch<IS>* EIG< IP, IS, OP, OS >::getNoveltySearch ( ) const

inline

Get the Novelty Search approach used in EIG.

Returns

◆ getPortfolio()

template<typename IP , typename IS , typename OP , typename OS >

vector< AbstractEA< OS > * > EIG< IP, IS, OP, OS >::getPortfolio

Get the portfolio of algorithms used in the evolutionary process.

Returns the portfolio used by the EIG.

Returns

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Returns

◆ getRepetitions()

template<typename IP , typename IS , typename OP , typename OS >

int EIG< IP, IS, OP, OS >::getRepetitions ( ) const

inline

Get the number of repetitions that each algorithm in the portfolio must perform over each instance at every generation of the EIG.

Returns: int

◆ getResults()

template<typename IP , typename IS , typename OP , typename OS >

Front< IS > EIG< IP, IS, OP, OS >::getResults

overridevirtual

Returns a set of individuals as a result of the evolutionary method. The individual are instances for the OP.

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Returns: Front of IS solutions

Implements AbstractEA< IS >.

Reimplemented in MapElites< IP, IS, OP, OS >.

◆ getSelection()

template<typename IP , typename IS , typename OP , typename OS >

Selection<IS>* EIG< IP, IS, OP, OS >::getSelection ( ) const

inline

Get the selection operator.

Returns: a raw pointer to Selection<IS>

◆ initProgress()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::initProgress

overrideprotectedvirtual

Initialises the number of generations performed to 0.

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Implements AbstractEA< IS >.

◆ isStoppingConditionReached()

template<typename IP , typename IS , typename OP , typename OS >

bool EIG< IP, IS, OP, OS >::isStoppingConditionReached

overrideprotectedvirtual

Checks whether the number of performed generations has reached the maximum allowed.

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Returns: True if the performedGenerations == maxGenerations, False otherwise

Implements AbstractEA< IS >.

◆ replacement()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::replacement ( vector< IS > & individuals )

protectedvirtual

Replacement method for the population. In this case, the replacement is a simple generational approach, which also includes the comparison of individuals with the archive and the feature set of the NS.

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Parameters

individuals New population of individuals to replace the current one

◆ reproduction()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::reproduction	(	IS &	solution1,
		IS &	solution2
	)

protectedvirtual

Executes the genetic operator over two instances to generate new individuals.

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Parameters

solution1	First child of type IS
solution2	Second child of type IS

◆ run()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::run

overridevirtual

Interface to run the EIG for generating a set of instances for the optimisation problem defined (OP)

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Implements AbstractEA< IS >.

Reimplemented in MapElites< IP, IS, OP, OS >.

◆ setCrossover()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::setCrossover ( unique_ptr< Crossover< IS >> cx )

inline

Set the crossover operator. Takes ownership of the pointer.

Parameters

cx

◆ setCrossoverRate()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::setCrossoverRate ( float cRate )

inline

Set the crossover rate used in EIG for applying the crossover operator to instances during the evolutionary process.

Parameters

cRate

◆ setInstanceFitness()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::setInstanceFitness ( unique_ptr< InstanceFitness > eval )

inline

Set the evaluation approach for the instances.

Parameters

eval

◆ setInstanceProblem()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::setInstanceProblem ( unique_ptr< IP > problem )

Set the Instance Generation Problem.

Sets the instance generation problem to toSolve.

Parameters

problem

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Parameters

problem

◆ setMutation()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::setMutation ( unique_ptr< Mutation< IS >> mut )

inline

Set the mutation operator. Takes ownership of the pointer.

Parameters

mut	unique_ptr

◆ setMutationRate()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::setMutationRate ( float mRate )

inline

Set the mutation rate used in EIG for mutation instances during the evolutionary process.

Parameters

mRate

◆ setNoveltySearch()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::setNoveltySearch ( unique_ptr< NoveltySearch< IS >> noveltySearch )

inline

Set the Novelty Search approach to use.

Parameters

noveltySearch

◆ setPortfolio()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::setPortfolio ( vector< unique_ptr< Alg >> & configs )

Set the portfolio of algorithms to use in the evolutionary process. The order of the algorithms in the portfolio is crucial. Notice that the first algorithm in the portfolio will be considered the target and therefore the instances will be generated according to its performance. Also, it is important to know the order of the remaining algorithms in order to analysis the results obtained since not tags are included in the results at this moment.

Updates the portfolio to solve the Instance Generation Problem.

Parameters

configs

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Parameters

configs

◆ setRepetitions()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::setRepetitions ( int reps )

inline

Set the number of repetitions that each algorithm in the portfolio must perform over each instance at every generation of the EIG. Due to stochastic nature of most algorithms used in EIG the number of repetitions should vary between 10 to 30. For deterministic heuristics can be set to one.

Parameters

reps

◆ setReplacement()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::setReplacement ( unique_ptr< Replacement< IS >> rep )

inline

Set the selection operator. Takes ownership of the pointer.

Parameters

sel

◆ setSelection()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::setSelection ( unique_ptr< Selection< IS >> sel )

inline

Set the selection operator. Takes ownership of the pointer.

Parameters

sel

◆ setWeightedFunction()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::setWeightedFunction	(	const float &	fW,
		const float &	nW
	)

inline

Set the weights to use in the WFF.

Parameters

fW
nW

◆ to_json()

template<typename IP , typename IS , typename OP , typename OS >

json EIG< IP, IS, OP, OS >::to_json

overridevirtual

Creates the JSON representation of the EIG.

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Returns

Reimplemented from AbstractEA< IS >.

Reimplemented in MapElites< IP, IS, OP, OS >.

◆ updateEvolution()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::updateEvolution ( vector< IS > & solutions )

protectedvirtual

Updates the evolution results of the algorithm for the given checkpoint.

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Parameters

solutions

◆ updateProgress()

template<typename IP , typename IS , typename OP , typename OS >

void EIG< IP, IS, OP, OS >::updateProgress

overrideprotectedvirtual

Updates the evolution by increasing the number of generations performed.

Template Parameters

IP	Type of Instance Problem to toSolve
IS	Type of instances of the Optimisation Problem
OP	Optimisation problem to toSolve with the configurations
OS	Solutions to the OP type

Implements AbstractEA< IS >.

Member Data Documentation

◆ algPortfolio

template<typename IP , typename IS , typename OP , typename OS >

vector<unique_ptr<Alg> > EIG< IP, IS, OP, OS >::algPortfolio

protected

Portfolio of algorithms to evaluate the instances

◆ averageFitness

template<typename IP , typename IS , typename OP , typename OS >

AverageFitness<OS> EIG< IP, IS, OP, OS >::averageFitness

protected

Avg fitness calculation for each instance at every generation

◆ bestFitness

template<typename IP , typename IS , typename OP , typename OS >

BestFitness<OS> EIG< IP, IS, OP, OS >::bestFitness

protected

Best fitness calculation for each algorithm after solving a instance at every generation

◆ crossover

template<typename IP , typename IS , typename OP , typename OS >

unique_ptr<Crossover<IS> > EIG< IP, IS, OP, OS >::crossover

protected

Crossover operator

◆ crossoverRate

template<typename IP , typename IS , typename OP , typename OS >

float EIG< IP, IS, OP, OS >::crossoverRate

protected

Crossover rate

◆ instProb

template<typename IP , typename IS , typename OP , typename OS >

unique_ptr<IP> EIG< IP, IS, OP, OS >::instProb

protected

Instance Generation Problem

◆ mutation

template<typename IP , typename IS , typename OP , typename OS >

unique_ptr<Mutation<IS> > EIG< IP, IS, OP, OS >::mutation

protected

Mutation operator

◆ mutationRate

template<typename IP , typename IS , typename OP , typename OS >

float EIG< IP, IS, OP, OS >::mutationRate

protected

Mutation rate

◆ noveltySearch

template<typename IP , typename IS , typename OP , typename OS >

unique_ptr<NoveltySearch<IS> > EIG< IP, IS, OP, OS >::noveltySearch

protected

Novelty Search

◆ repetitions

template<typename IP , typename IS , typename OP , typename OS >

int EIG< IP, IS, OP, OS >::repetitions

protected

Number of repetitions to perform by the algorithms

◆ selection

template<typename IP , typename IS , typename OP , typename OS >

unique_ptr<Selection<IS> > EIG< IP, IS, OP, OS >::selection

protected

Selection operator

The documentation for this class was generated from the following file:

include/dignea/generator/EIG.h

Public Types

Public Member Functions

Protected Member Functions

Protected Attributes

Additional Inherited Members

Detailed Description

template<typename IP, typename IS, typename OP, typename OS> class EIG< IP, IS, OP, OS >

Member Typedef Documentation

◆ Alg

Constructor & Destructor Documentation

◆ EIG()

Member Function Documentation

◆ createInitialPopulation()

◆ evaluateIndividual()

◆ evaluatePopulation()

◆ evaluationPhase()

◆ finishProgress()

◆ getCrossover()

◆ getCrossoverRate()

◆ getEvaluation()

◆ getGenerations()

◆ getInstanceProblem()

◆ getMutation()

◆ getMutationRate()

◆ getName()

◆ getNoveltySearch()

◆ getPortfolio()

◆ getRepetitions()

◆ getResults()

◆ getSelection()

◆ initProgress()

◆ isStoppingConditionReached()

◆ replacement()

◆ reproduction()

◆ run()

◆ setCrossover()

◆ setCrossoverRate()

◆ setInstanceFitness()

◆ setInstanceProblem()

◆ setMutation()

◆ setMutationRate()

◆ setNoveltySearch()

◆ setPortfolio()

◆ setRepetitions()

◆ setReplacement()

◆ setSelection()

◆ setWeightedFunction()

◆ to_json()

◆ updateEvolution()

◆ updateProgress()

Member Data Documentation

◆ algPortfolio

◆ averageFitness

◆ bestFitness

◆ crossover

◆ crossoverRate

◆ instProb

◆ mutation

◆ mutationRate

◆ noveltySearch

◆ repetitions

◆ selection

template<typename IP, typename IS, typename OP, typename OS>
class EIG< IP, IS, OP, OS >