Control object for one iteration of adaptive importance sampling. More...

#include <AdaptiveImportanceSampler.h>

Public Member Functions
	ImportanceSamplerControl ()

Public Attributes
int	nSamples
	"Number of Monte Carlo samples to draw" ; More...

int	nUpdateSteps
	"Number of Expectation-Maximization update iterations" ; More...

double	tau1
	"Damping parameter for E-M update (see Mixture::updateEM)" ; More...

double	tau2
	"Damping parameter for E-M update (see Mixture::updateEM)" ; More...

double	targetPerplexity
	"Minimum value for normalized perplexity after this iteration; if the actual value is less " "than this, this iteration will be repeated up to maxRepeat times until the target is met. " "In addition, if any previous iteration meets this target, this iteration will be skipped." ; More...

int	maxRepeat
	"Maximum number of times this iteration will be repeated to meet the perplexityTarget" ; More...

Detailed Description

Control object for one iteration of adaptive importance sampling.

See also: AdaptiveImportanceSampler, AdaptiveImportanceSamplerTask

Definition at line 41 of file AdaptiveImportanceSampler.h.

Constructor & Destructor Documentation

◆ ImportanceSamplerControl()

lsst::meas::modelfit::ImportanceSamplerControl::ImportanceSamplerControl ( )

inline

Definition at line 58 of file AdaptiveImportanceSampler.h.

                                :
         nSamples(2000), nUpdateSteps(2), tau1(1E-4), tau2(0.5), targetPerplexity(1.0), maxRepeat(0)
     {}

Member Data Documentation

◆ maxRepeat

int lsst::meas::modelfit::ImportanceSamplerControl::maxRepeat

"Maximum number of times this iteration will be repeated to meet the perplexityTarget" ;

Definition at line 56 of file AdaptiveImportanceSampler.h.

◆ nSamples

int lsst::meas::modelfit::ImportanceSamplerControl::nSamples

"Number of Monte Carlo samples to draw" ;

Definition at line 43 of file AdaptiveImportanceSampler.h.

◆ nUpdateSteps

int lsst::meas::modelfit::ImportanceSamplerControl::nUpdateSteps

"Number of Expectation-Maximization update iterations" ;

Definition at line 44 of file AdaptiveImportanceSampler.h.

◆ targetPerplexity

double lsst::meas::modelfit::ImportanceSamplerControl::targetPerplexity

"Minimum value for normalized perplexity after this iteration; if the actual value is less " "than this, this iteration will be repeated up to maxRepeat times until the target is met. " "In addition, if any previous iteration meets this target, this iteration will be skipped." ;

Definition at line 52 of file AdaptiveImportanceSampler.h.