This class represents a function with constant random positive (semi) definite derivative. More...

#include <random_positive_definite.h>

Inheritance diagram for highprecisionpso::RandomPositiveDefiniteSecondDerivative:

Public Member Functions
	RandomPositiveDefiniteSecondDerivative (std::vector< std::string > rng_description)
	The constructor, where the description of the random number generator can be specified. More...

	~RandomPositiveDefiniteSecondDerivative ()
	The destructor. More...

mpf_t *	DistanceTo1DLocalOptimum (const std::vector< mpf_t * > &pos, int d)
	Calculates for a position and a dimension the next local optimum if only the value of the position in the specified dimension is varied and returns the distance to it. More...

mpf_t *	Eval (const std::vector< mpf_t * > &pos)
	Returns the product transpose(pos) * A * pos, where A is a random matrix. More...

std::string	GetName ()
	Returns a string representation of this object. More...

Public Member Functions inherited from highprecisionpso::Function
	Function ()
	A constructor. More...

	Function (double search_space_lower_bound, double search_space_upper_bound)
	A constructor, specifying the lower and upper search space bounds. More...

	Function (std::vector< double > search_space_lower_bound, std::vector< double > search_space_upper_bound)
	A constructor specifying the lower and upper search space bounds. More...

virtual	~Function ()
	The destructor. More...

virtual mpf_t *	DistanceTo1DLocalOptimumTernarySearch (const std::vector< mpf_t * > &pos, int d, mpf_t *start_distance)
	Calculates for a position and a dimension the next local optimum if only the value of the position in the specified dimension is varied and returns the distance to it. More...

mpf_t *	Evaluate (const std::vector< mpf_t * > &pos)
	This is the initial function for evaluation of a position. More...

virtual std::vector< mpf_t * >	GetLowerSearchSpaceBound ()
	Returns the lower search space bound of the search space for this objective function. More...

virtual std::vector< mpf_t * >	GetUpperSearchSpaceBound ()
	Returns the upper search space bound of the search space for this objective function. More...

void	InitSearchSpaceBounds ()
	Initializes the search space bounds. More...

void	SetFunctionBound (SearchSpaceBound *search_space_bound)
	Updates the bound of the search space according to the given data. More...

Detailed Description

This class represents a function with constant random positive (semi) definite derivative.

Constructor & Destructor Documentation

highprecisionpso::RandomPositiveDefiniteSecondDerivative::RandomPositiveDefiniteSecondDerivative ( std::vector< std::string > rng_description )

The constructor, where the description of the random number generator can be specified.

Parameters

rng_description The description of the random number generator.

highprecisionpso::RandomPositiveDefiniteSecondDerivative::~RandomPositiveDefiniteSecondDerivative ( )

The destructor.

Member Function Documentation

mpf_t* highprecisionpso::RandomPositiveDefiniteSecondDerivative::DistanceTo1DLocalOptimum	(	const std::vector< mpf_t * > &	pos,
		int	d
	)

virtual

Calculates for a position and a dimension the next local optimum if only the value of the position in the specified dimension is varied and returns the distance to it.

For a given position all values (beside the value in the specified dimension) are regarded as constants. By the evaluation of a quadratic equation, where the value of the position in the specified dimension is varied, a local optimum is calculated. The result is the distance of the evaluated position and the found local optimum.

Parameters

pos	The position.
d	The index of the observed dimension.

Returns: The distance to the next local optimum in a single dimension setting.

Reimplemented from highprecisionpso::Function.

mpf_t* highprecisionpso::RandomPositiveDefiniteSecondDerivative::Eval ( const std::vector< mpf_t * > & pos )

virtual

Returns the product transpose(pos) * A * pos, where A is a random matrix.

The random matrix A is generated according to the specified random number generator. Let D be the number of dimensions. Then a D times D matrix of random long long values is generated. This matrix is multiplied by its transposed matrix to generate the positive (semi) definite matrix A. If the random number generator is not bad then the probability, that the matrix A is positive definite and not only positive semi definite, is very large.

Parameters

pos	The position vector.

Returns: Transpose(pos) * A * pos.

Implements highprecisionpso::Function.

std::string highprecisionpso::RandomPositiveDefiniteSecondDerivative::GetName ( )

virtual

Returns a string representation of this object.

Returns: The name of the object.

Implements highprecisionpso::Function.

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

src/function/random_positive_definite.h

Public Member Functions

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation