Surrogate base class, to be derived in order to create strategy to be used along with CMA-ES. More...

#include <libcmaes/surrogatestrategy.h>

Inheritance diagram for libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >:

Public Member Functions
	SurrogateStrategy (FitFunc &func, CMAParameters< TGenoPheno > &parameters)
	constructor

int	train (const std::vector< Candidate > &candidates, const dMat &cov)
	train a surrogate model

int	predict (std::vector< Candidate > &candidates, const dMat &cov)
	predict from a surrogate model

double	compute_error (const std::vector< Candidate > &test_set, const dMat &cov=dMat(0, 0))
	compute surrogate model error (copies and sorts the test_set)

bool	do_train () const
	conditionals on training, to be specialized in inherited surrogate strategies

void	set_ftrain (const CSurrFunc &train)
	sets the training function

void	set_fpredict (const SurrFunc &predict)
	sets the prediction function

void	set_l (const int &l)
	sets the size of the training set (number of points)

int	get_l () const
	gets the size of the training set (number of points)

void	set_exploit (const bool &exploit)
	sets whether to exploit the surrogate model

bool	get_exploit () const
	gets the state of surrogate model exploitation

double	get_train_error () const
	returns the surrogate model training error

double	get_test_error () const
	returns the surrogate model test error

void	set_train_error (const double &err)
	sets training error

void	set_test_error (const double &err)
	sets the test error and updates the smoothed test err.

void	add_to_training_set (const Candidate &c)
	adds a point to the training set (candidate = points + objective function value)

void	set_nsteps (const int &nsteps)
	sets the lifelength of the surrogate, i.e. the number of steps in between to training steps

void	reset_training_set ()
	resets training set and related information, useful when using algorithms with restarts

int	get_nsteps () const
	returns the current surrogate lifelength

Protected Attributes
bool	_exploit = true

int	_l = 200

std::vector< Candidate >	_tset

CSurrFunc	_train

SurrFunc	_predict

double	_train_err = 0.0

double	_test_err = 0.0

double	_smooth_test_err = 0.5

double	_beta_err = 0.2

int	_nsteps = 1

int	_auto_nsteps = false

Detailed Description

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>
class libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >

Surrogate base class, to be derived in order to create strategy to be used along with CMA-ES.

Constructor & Destructor Documentation

◆ SurrogateStrategy()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate , class TGenoPheno >

libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::SurrogateStrategy	(	FitFunc &	func,
		CMAParameters< TGenoPheno > &	parameters
	)

constructor

Parameters

func	objective function to minimize
parameters	optimization parameters

Member Function Documentation

◆ add_to_training_set()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate , class TGenoPheno >

void libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::add_to_training_set ( const Candidate & c )

adds a point to the training set (candidate = points + objective function value)

Parameters

c	point to add to the training set

◆ compute_error()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate , class TGenoPheno >

double libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::compute_error	(	const std::vector< Candidate > &	test_set,
		const dMat &	cov = `dMat(0,0)`
	)

compute surrogate model error (copies and sorts the test_set)

Parameters

test_set	the candidate points along with their objective function values for model evaluation
cov	possibly empty covariance matrix in order to re-scale the points before error estimation

Returns: surrogate model error estimate

◆ do_train()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

bool libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::do_train ( ) const

inline

conditionals on training, to be specialized in inherited surrogate strategies

Returns: whether to train surrogate

◆ get_exploit()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

bool libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::get_exploit ( ) const

inline

gets the state of surrogate model exploitation

Returns: whether the surrogate model is being exploited

◆ get_l()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

int libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::get_l ( ) const

inline

gets the size of the training set (number of points)

Returns: size of the training set

◆ get_nsteps()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

int libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::get_nsteps ( ) const

inline

returns the current surrogate lifelength

Returns: current surrogate lifelength

◆ get_test_error()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

double libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::get_test_error ( ) const

inline

returns the surrogate model test error

Returns: surrogate model test error

◆ get_train_error()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

double libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::get_train_error ( ) const

inline

returns the surrogate model training error

Returns: surrogate model training error

◆ predict()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

int libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::predict	(	std::vector< Candidate > &	candidates,
		const dMat &	cov
	)

inline

predict from a surrogate model

Parameters

candidates	set of points for which value is to be predicted
cov	a possibly empty covariance matrix in order to re-scale points before predicting

Returns: prediction status

◆ set_exploit()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

void libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::set_exploit ( const bool & exploit )

inline

sets whether to exploit the surrogate model

Parameters

exploit whether to exploit the surrogate model

◆ set_fpredict()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

void libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::set_fpredict ( const SurrFunc & predict )

inline

sets the prediction function

Parameters

prediction function

◆ set_ftrain()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

void libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::set_ftrain ( const CSurrFunc & train )

inline

sets the training function

Parameters

training function

◆ set_l()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

void libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::set_l ( const int & l )

inline

sets the size of the training set (number of points)

Parameters

l	size of the training set

◆ set_nsteps()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

void libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::set_nsteps ( const int & nsteps )

inline

sets the lifelength of the surrogate, i.e. the number of steps in between to training steps

Parameters

nsteps surrogate lifelength, -1 for automatic determination

◆ set_test_error()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate , class TGenoPheno >

void libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::set_test_error ( const double & err )

sets the test error and updates the smoothed test err.

Parameters

err	test error

◆ set_train_error()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

void libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::set_train_error ( const double & err )

inline

sets training error

Parameters

err	training error

◆ train()

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

int libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::train	(	const std::vector< Candidate > &	candidates,
		const dMat &	cov
	)

inline

train a surrogate model

Parameters

candidates	set of points along with objective function value
cov	a possibly empty covariance matrix in order to re-scale points before training

Returns: training status

Member Data Documentation

◆ _auto_nsteps

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

int libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::_auto_nsteps = false

protected

whether to automatically set the surrogate lifelength.

◆ _beta_err

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

double libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::_beta_err = 0.2

protected

smoothing constant.

◆ _exploit

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

bool libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::_exploit = true

protected

whether to exploit or test the surrogate.

◆ _l

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

int libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::_l = 200

protected

number of training samples. set to floor(30*sqrt(n)) in constructor.

◆ _nsteps

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

int libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::_nsteps = 1

protected

steps in between two training phases.

◆ _predict

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

SurrFunc libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::_predict

protected

custom prediction function.

◆ _smooth_test_err

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

double libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::_smooth_test_err = 0.5

protected

smoothed test error as (1-\beta_err)*_test_err + \beta_err * new_test_err

◆ _test_err

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

double libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::_test_err = 0.0

protected

current surrogate model error estimate.

◆ _train

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

CSurrFunc libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::_train

protected

custom training function.

◆ _train_err

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

double libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::_train_err = 0.0

protected

current surrogate training error.

◆ _tset

template<template< class U, class V > class TStrategy, class TCovarianceUpdate = CovarianceUpdate, class TGenoPheno = GenoPheno<NoBoundStrategy>>

std::vector<Candidate> libcmaes::SurrogateStrategy< TStrategy, TCovarianceUpdate, TGenoPheno >::_tset

protected

current training set.

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

include/libcmaes/surrogatestrategy.h
src/surrogatestrategy.cc

Public Member Functions

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

◆ SurrogateStrategy()

Member Function Documentation

◆ add_to_training_set()

◆ compute_error()

◆ do_train()

◆ get_exploit()

◆ get_l()

◆ get_nsteps()

◆ get_test_error()

◆ get_train_error()

◆ predict()

◆ set_exploit()

◆ set_fpredict()

◆ set_ftrain()

◆ set_l()

◆ set_nsteps()

◆ set_test_error()

◆ set_train_error()

◆ train()

Member Data Documentation

◆ _auto_nsteps

◆ _beta_err

◆ _exploit

◆ _l

◆ _nsteps

◆ _predict

◆ _smooth_test_err

◆ _test_err

◆ _train

◆ _train_err

◆ _tset