Abstract base class for objective functions to be minimized. More...

#include <FitnessFunction.h>

Inheritance diagram for galib::FitnessFunction< GeneType >:

Public Member Functions
virtual	~FitnessFunction ()=default

virtual double	evaluate (const std::vector< GeneType > &phenotype) const =0
	Evaluates the fitness of a candidate solution.

virtual std::size_t	size () const =0
	Returns the dimensionality of the optimization problem.

virtual std::string	name () const
	Short identifier used to select a GPU evaluation kernel.

virtual GeneType	getLowerBound (std::size_t dimension) const =0
	Gets the lower bound for a specific dimension.

GeneType	getLowerBound () const
	Convenience method to get the lower bound of the first dimension.

virtual GeneType	getUpperBound (std::size_t dimension) const =0
	Gets the upper bound for a specific dimension.

GeneType	getUpperBound () const
	Convenience method to get the upper bound of the first dimension.

Detailed Description

template<typename GeneType = double>
class galib::FitnessFunction< GeneType >

Abstract base class for objective functions to be minimized.

Users should inherit from this class to define their own optimization problems, or use FunctionalFitness to wrap a lambda.

Template Parameters

GeneType The numeric type of the genes (default: double).

Definition at line 21 of file FitnessFunction.h.

Constructor & Destructor Documentation

◆ ~FitnessFunction()

template<typename GeneType = double>

virtual galib::FitnessFunction< GeneType >::~FitnessFunction ( )

virtualdefault

Member Function Documentation

◆ evaluate()

template<typename GeneType = double>

virtual double galib::FitnessFunction< GeneType >::evaluate ( const std::vector< GeneType > & phenotype ) const

pure virtual

Evaluates the fitness of a candidate solution.

Lower return values indicate better fitness (minimization convention).

Parameters

phenotype Gene values of the individual to evaluate.

Returns: Fitness score.

Implemented in galib::benchmark::RastriginFunction< GeneType >, galib::benchmark::DeJongF5Function< GeneType >, galib::benchmark::HeavyTrigFunction< GeneType >, galib::benchmark::HimmelblauFitness< GeneType >, galib::benchmark::SphereFunction< GeneType >, and galib::FunctionalFitness< GeneType >.

Referenced by galib::DifferentialEvolutionGA< GeneType >::run().

◆ getLowerBound() [1/2]

template<typename GeneType = double>

GeneType galib::FitnessFunction< GeneType >::getLowerBound ( ) const

inline

Convenience method to get the lower bound of the first dimension.

Returns: Lower bound of dimension 0.

Definition at line 63 of file FitnessFunction.h.

References galib::FitnessFunction< GeneType >::getLowerBound().

Referenced by galib::FitnessFunction< GeneType >::getLowerBound().

◆ getLowerBound() [2/2]

template<typename GeneType = double>

virtual GeneType galib::FitnessFunction< GeneType >::getLowerBound ( std::size_t dimension ) const

pure virtual

Gets the lower bound for a specific dimension.

Parameters

dimension Index of the dimension.

Returns: Lower bound value.

Implemented in galib::benchmark::HimmelblauFitness< GeneType >, galib::benchmark::DeJongF5Function< GeneType >, galib::FunctionalFitness< GeneType >, galib::benchmark::HeavyTrigFunction< GeneType >, galib::benchmark::RastriginFunction< GeneType >, and galib::benchmark::SphereFunction< GeneType >.

Referenced by galib::utils::AlgorithmBuilder< GeneType >::buildCellularGA(), galib::utils::AlgorithmBuilder< GeneType >::buildIslandGA(), galib::utils::AlgorithmBuilder< GeneType >::buildStandardGA(), and galib::DifferentialEvolutionGA< GeneType >::run().

◆ getUpperBound() [1/2]

template<typename GeneType = double>

GeneType galib::FitnessFunction< GeneType >::getUpperBound ( ) const

inline

Convenience method to get the upper bound of the first dimension.

Returns: Upper bound of dimension 0.

Definition at line 78 of file FitnessFunction.h.

References galib::FitnessFunction< GeneType >::getUpperBound().

Referenced by galib::FitnessFunction< GeneType >::getUpperBound().

◆ getUpperBound() [2/2]

template<typename GeneType = double>

virtual GeneType galib::FitnessFunction< GeneType >::getUpperBound ( std::size_t dimension ) const

pure virtual

Gets the upper bound for a specific dimension.

Parameters

dimension Index of the dimension.

Returns: Upper bound value.

Implemented in galib::benchmark::HimmelblauFitness< GeneType >, galib::benchmark::DeJongF5Function< GeneType >, galib::FunctionalFitness< GeneType >, galib::benchmark::HeavyTrigFunction< GeneType >, galib::benchmark::RastriginFunction< GeneType >, and galib::benchmark::SphereFunction< GeneType >.

Referenced by galib::utils::AlgorithmBuilder< GeneType >::buildCellularGA(), galib::utils::AlgorithmBuilder< GeneType >::buildIslandGA(), galib::utils::AlgorithmBuilder< GeneType >::buildStandardGA(), and galib::DifferentialEvolutionGA< GeneType >::run().

◆ name()

template<typename GeneType = double>

virtual std::string galib::FitnessFunction< GeneType >::name ( ) const

inlinevirtual

Short identifier used to select a GPU evaluation kernel.

Override in benchmark subclasses to enable GPU-side evaluation ("Sphere", "Rastrigin", "HeavyTrig"). Return empty string (default) to use the CPU callback path.