CMAES最適化のデスクリプタ

#include <SprUTOptimizer.h>

Public 型
enum	Weights { UTCMA_UNINITIALIZED_WEIGHTS, UTCMA_LINEAR_WEIGHTS, UTCMA_EQUAL_WEIGHTS, UTCMA_LOG_WEIGHTS }
	Determines the method used to initialize the weights.

Public メソッド
	SPR_DESCDEF (UTCMAESOptimizer)

Public 変数
double	stopMaxFunEvals
	Maximal number of objective function evaluations.

double	facmaxeval

double	stopMaxIter
	Maximal number of iterations.

struct {
bool flg

double val

}	stStopFitness
	Minimal fitness value. Only activated if flg is true.

double	stopTolFun
	Minimal value difference.

double	stopTolFunHist
	Minimal history value difference.

double	stopTolX
	Minimal search space step size.

double	stopTolUpXFactor
	Defines the maximal condition number.

int	lambda
	Population size. [詳細]

int	mu
	Number of individuals used to recompute the mean.

double	mucov

double	mueff
	Variance effective selection mass, should be lambda/4.

std::vector< double >	weights
	Weights used to recombinate the mean sum up to one.

double	damps
	Damping parameter for step-size adaption, d = inifinity or 0 means adaption is turned off, usually close to one.

double	cs
	cs^-1 (approx. [詳細]

double	ccumcov

double	ccov
	ccov^-1 (approx. [詳細]

double	diagonalCov

struct {
double modulo

double maxtime

}	updateCmode

double	facupdateCmode

enum Spr::UTCMAESOptimizerDesc::Weights	weightMode

変数

double ccov

n/4) is the backward time horizon for the evolution path pc and larger than one.

double cs

n/3) is the backward time horizon for the evolution path ps and larger than one.

int lambda

Number of samples per iteration, at least two, generally > 4.

この構造体の説明は次のファイルから生成されました:

Public 型