Holder of the set of evolving solutions from running an instance of CMA-ES. More...

#include <cmasolutions.h>

Public Member Functions
	CMASolutions ()
	dummy constructor.

template<class TGenoPheno = GenoPheno<NoBoundStrategy>>
	CMASolutions (Parameters< TGenoPheno > &p)
	initializes solutions from stochastic optimization parameters. More...

void	sort_candidates ()
	sorts the current internal set of solution candidates.

void	update_best_candidates ()
	updates the history of best candidates, as well as other meaningful values, typically used in termination criteria. More...

void	update_eigenv (const dVec &eigenvalues, const dMat &eigenvectors)
	updates reference eigenvalue and eigenvectors, for use in termination criteria. More...

Candidate	best_candidate () const
	returns current best solution candidate. NOTE: candidates MUST be sorted More...

Candidate	get_best_seen_candidate () const
	returns the best seen candidate. More...

Candidate	get_worst_seen_candidate () const
	returns the worst seen candidate. More...

Candidate &	get_candidate (const int &r)
	get a reference to the r-th candidate in current set More...

Candidate	get_candidate (const int &r) const

std::vector< Candidate > &	candidates ()
	get a reference to the full candidate set

int	size () const
	number of candidate solutions. More...

void	reset ()
	resets the solution object in order to restart from the current solution with fresh covariance matrix. Note: experimental.

void	reset_as_fixed (const int &k)
	re-arrange solution object such that parameter 'k' is fixed (i.e. removed). More...

bool	get_pli (const int &k, pli &p) const
	get profile likelihood if previously computed.

int	dim () const
	return problem dimension. More...

double	edm () const
	returns expected distance to minimum. More...

dMat	cov () const
	returns error covariance matrix More...

const dMat &	cov_ref () const
	returns reference to error covariance matrix More...

const double *	cov_data () const
	returns pointer to covariance matrix array More...

dMat	full_cov () const
	returns full covariance matrix. Similar to cov() but in case of linear-sized algorithms like sep and vd, returns the full covariance matrix anyways. More...

dMat	sepcov () const
	returns separable covariance diagonal matrix, only applicable to sep-CMA-ES algorithms. More...

const dMat &	sepcov_ref () const
	returns reference to separable covariance diagonal vector, only applicable to sep-CMA-ES algorithms. More...

const double *	sepcov_data () const
	returns pointer to covariance diagnoal vector More...

dMat	csqinv () const
	returns inverse root square of covariance matrix More...

dMat	sepcsqinv () const
	returns inverse root square of separable covariance diagonal matrix, only applicable to sep-CMA-ES algorithms. More...

template<class TGenoPheno = GenoPheno<NoBoundStrategy>>
dVec	stds (const CMAParameters< TGenoPheno > &cmaparams) const

template<class TGenoPheno = GenoPheno<NoBoundStrategy>>
dVec	errors (const CMAParameters< TGenoPheno > &cmaparams) const

dMat	corr () const
	returns correlation matrix More...

double	corr (const int &i, const int &j) const
	returns correlation between parameter i and j (useful in large-scale settings) More...

double	sigma () const
	returns current value of step-size sigma More...

dVec	xmean () const
	returns current distribution's mean in parameter space More...

void	set_xmean (const dVec &xmean)
	sets the current distributions' mean in parameter space More...

int	run_status () const
	returns current optimization status. More...

std::string	status_msg () const
	returns current optimization status' message. More...

int	elapsed_time () const
	returns currently elapsed time spent on optimization More...

int	elapsed_last_iter () const
	returns time spent on last iteration More...

int	niter () const
	returns current number of iterations More...

int	nevals () const
	returns current budget (number of objective function calls) More...

double	min_eigenv () const
	returns current minimal eigen value More...

double	max_eigenv () const
	returns current maximal eigen value More...

bool	updated_eigen () const
	returns whether the last update is lazy More...

int	fevals () const
	returns current number of objective function evaluations More...

dVec	eigenvalues () const
	returns last computed eigenvalues More...

dMat	eigenvectors () const
	returns last computed eigenvectors More...

template<class TGenoPheno = GenoPheno<NoBoundStrategy>>
std::ostream &	print (std::ostream &out, const int &verb_level=0, const TGenoPheno &gp=TGenoPheno()) const
	print the solution object out. More...

Friends
template<class U , class V >
class	CMAStrategy

template<class U , class V , class W >
class	ESOptimizer

template<class U , class V , class W >
class	ESOStrategy

template<class U >
class	CMAStopCriteria

template<class U , class V >
class	IPOPCMAStrategy

template<class U , class V >
class	BIPOPCMAStrategy

class	CovarianceUpdate

class	ACovarianceUpdate

template<class U >
class	errstats

class	VDCMAUpdate

Detailed Description

Holder of the set of evolving solutions from running an instance of CMA-ES.

Constructor & Destructor Documentation

template<class TGenoPheno >

libcmaes::CMASolutions::CMASolutions ( Parameters< TGenoPheno > & p )

initializes solutions from stochastic optimization parameters.

Parameters

p	parameters

Member Function Documentation

Candidate libcmaes::CMASolutions::best_candidate ( ) const

inline

returns current best solution candidate. NOTE: candidates MUST be sorted

Returns: current best candidate

See also: CMASolutions::sort_candidates

dMat libcmaes::CMASolutions::corr ( ) const

returns correlation matrix

Returns: correlation matrix

double libcmaes::CMASolutions::corr	(	const int &	i,
		const int &	j
	)		const

returns correlation between parameter i and j (useful in large-scale settings)

Returns: correlation between parameter i and j

dMat libcmaes::CMASolutions::cov ( ) const

inline

returns error covariance matrix

Returns: error covariance matrix

const double* libcmaes::CMASolutions::cov_data ( ) const

inline

returns pointer to covariance matrix array

Returns: pointer to covariance matrix array

const dMat& libcmaes::CMASolutions::cov_ref ( ) const

inline

returns reference to error covariance matrix

Returns: error covariance matrix

dMat libcmaes::CMASolutions::csqinv ( ) const

inline

returns inverse root square of covariance matrix

Returns: square root of error covariance matrix

int libcmaes::CMASolutions::dim ( ) const

inline

return problem dimension.

Returns: problem dimension

double libcmaes::CMASolutions::edm ( ) const

inline

returns expected distance to minimum.

Returns: edm

dVec libcmaes::CMASolutions::eigenvalues ( ) const

inline

returns last computed eigenvalues

Returns: last computed eigenvalues

dMat libcmaes::CMASolutions::eigenvectors ( ) const

inline

returns last computed eigenvectors

Returns: last computed eigenvectors

int libcmaes::CMASolutions::elapsed_last_iter ( ) const

inline

returns time spent on last iteration

Returns: time spent on last iteration

int libcmaes::CMASolutions::elapsed_time ( ) const

inline

returns currently elapsed time spent on optimization

Returns: time spent on optimization

template<class TGenoPheno = GenoPheno<NoBoundStrategy>>

dVec libcmaes::CMASolutions::errors ( const CMAParameters< TGenoPheno > & cmaparams ) const

inline

Returns: standard deviation vector

Parameters

cmaparams parameter object that hold the genotype/phenotype transform

Returns: standard deviation vector

int libcmaes::CMASolutions::fevals ( ) const

inline

returns current number of objective function evaluations

Returns: number of objective function evaluations

dMat libcmaes::CMASolutions::full_cov ( ) const

returns full covariance matrix. Similar to cov() but in case of linear-sized algorithms like sep and vd, returns the full covariance matrix anyways.

Returns: full size covariance matrix

Candidate libcmaes::CMASolutions::get_best_seen_candidate ( ) const

inline

returns the best seen candidate.

Returns: best seen candidate

Candidate& libcmaes::CMASolutions::get_candidate ( const int & r )

inline

get a reference to the r-th candidate in current set

Parameters

r	candidate position

Candidate libcmaes::CMASolutions::get_worst_seen_candidate ( ) const

inline

returns the worst seen candidate.

Returns: worst seen candidate

double libcmaes::CMASolutions::max_eigenv ( ) const

inline

returns current maximal eigen value

Returns: maximal eigen value

double libcmaes::CMASolutions::min_eigenv ( ) const

inline

returns current minimal eigen value

Returns: minimal eigen value

int libcmaes::CMASolutions::nevals ( ) const

inline

returns current budget (number of objective function calls)

Returns: number of objective function calls

int libcmaes::CMASolutions::niter ( ) const

inline

returns current number of iterations

Returns: number of iterations

template<class TGenoPheno >

template std::ostream & libcmaes::CMASolutions::print	(	std::ostream &	out,
		const int &	verb_level = `0`,
		const TGenoPheno &	gp = `TGenoPheno()`
	)		const

print the solution object out.

Parameters

out	output stream
verb_level	verbosity level: 0 for short, 1 for debug.

void libcmaes::CMASolutions::reset_as_fixed ( const int & k )

re-arrange solution object such that parameter 'k' is fixed (i.e. removed).

Parameters

k	index of the parameter to remove.

int libcmaes::CMASolutions::run_status ( ) const

inline

returns current optimization status.

Returns: status

dMat libcmaes::CMASolutions::sepcov ( ) const

inline

returns separable covariance diagonal matrix, only applicable to sep-CMA-ES algorithms.

Returns: error covariance diagonal vector

const double* libcmaes::CMASolutions::sepcov_data ( ) const

inline

returns pointer to covariance diagnoal vector

Returns: pointer to covariance diagonal array

const dMat& libcmaes::CMASolutions::sepcov_ref ( ) const

inline

returns reference to separable covariance diagonal vector, only applicable to sep-CMA-ES algorithms.

Returns: error covariance diagonal vector

dMat libcmaes::CMASolutions::sepcsqinv ( ) const

inline

returns inverse root square of separable covariance diagonal matrix, only applicable to sep-CMA-ES algorithms.

Returns: square root of error covariance diagonal matrix

void libcmaes::CMASolutions::set_xmean ( const dVec & xmean )

inline

sets the current distributions' mean in parameter space

Parameters

xmean mean vector

double libcmaes::CMASolutions::sigma ( ) const

inline

returns current value of step-size sigma

Returns: current step-size

int libcmaes::CMASolutions::size ( ) const

inline

number of candidate solutions.

Returns: current number of solution candidates.

std::string libcmaes::CMASolutions::status_msg ( ) const

inline

returns current optimization status' message.

Returns: status message

template<class TGenoPheno = GenoPheno<NoBoundStrategy>>

dVec libcmaes::CMASolutions::stds ( const CMAParameters< TGenoPheno > & cmaparams ) const

inline

Returns: the unscaled standard deviation vector Note: this is only useful to compare amond standard deviations To get the true rescaled estimate of the error, use errors()

Parameters

cmaparams parameter object that hold the genotype/phenotype transform

Returns: unscaled standard deviation vector

void libcmaes::CMASolutions::update_best_candidates ( )

updates the history of best candidates, as well as other meaningful values, typically used in termination criteria.

See also: CMAStopCriteria

void libcmaes::CMASolutions::update_eigenv	(	const dVec &	eigenvalues,
		const dMat &	eigenvectors
	)

updates reference eigenvalue and eigenvectors, for use in termination criteria.

See also: CMAStopCriteria

bool libcmaes::CMASolutions::updated_eigen ( ) const

inline

returns whether the last update is lazy

Returns: whether the last update is lazy

dVec libcmaes::CMASolutions::xmean ( ) const

inline

returns current distribution's mean in parameter space

Returns: mean

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

src/cmasolutions.h
src/cmasolutions.cc

Public Member Functions

Friends

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation