SG++-Doxygen-Documentation
Public Attributes | List of all members
sgpp::base::AdaptivityConfiguration Struct Reference

structure that can be used by application to define adaptivity strategies More...

#include <Grid.hpp>

Public Attributes

bool errorBasedRefinement = false
 other refinement strategy, that is more expensive, but yields better results More...
 
size_t errorBufferSize = 3
 amount of error values to consider when checking for convergence in case of error based refinement More...
 
double errorConvergenceThreshold = 0.001
 threshold for convergence in case error based refinement is applied More...
 
size_t errorMinInterval = 0
 minimum amount of iterations before the next refinement is allowed to happen in case of error based refinement More...
 
bool levelPenalize = false
 determines if finer grid levels should be penalized when finding points to refine More...
 
bool maxLevelType_
 refinement type: false: classic, true: maxLevel More...
 
size_t noPoints_
 max. number of points to be refined More...
 
size_t numRefinements_
 number of refinements More...
 
double percent_ = 1.0
 max. percent of points to be refined More...
 
bool precomputeEvaluations = true
 in case of zero corssing based refinement: determines if evaluations should be precomupted More...
 
RefinementFunctorType refinementFunctorType = RefinementFunctorType::Surplus
 refinement indicator More...
 
size_t refinementPeriod = 1
 refinement will be triggered each refinementPeriod instances (approximately) in case of non error based refinement More...
 
std::vector< double > scalingCoefficients = std::vector<double>()
 in case of data based refinements: determines the scaling coefficients for each class More...
 
double threshold_
 refinement threshold for surpluses More...
 

Detailed Description

structure that can be used by application to define adaptivity strategies

Member Data Documentation

◆ errorBasedRefinement

bool sgpp::base::AdaptivityConfiguration::errorBasedRefinement = false

other refinement strategy, that is more expensive, but yields better results

Referenced by sgpp::datadriven::RefinementMonitorFactory::createRefinementMonitor(), sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig(), sgpp::datadriven::FitterConfigurationDensityEstimation::setupDefaults(), and sgpp::datadriven::LearnerBase::train().

◆ errorBufferSize

size_t sgpp::base::AdaptivityConfiguration::errorBufferSize = 3

amount of error values to consider when checking for convergence in case of error based refinement

Referenced by sgpp::datadriven::RefinementMonitorFactory::createRefinementMonitor(), and sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig().

◆ errorConvergenceThreshold

double sgpp::base::AdaptivityConfiguration::errorConvergenceThreshold = 0.001

threshold for convergence in case error based refinement is applied

Referenced by sgpp::datadriven::RefinementMonitorFactory::createRefinementMonitor(), and sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig().

◆ errorMinInterval

size_t sgpp::base::AdaptivityConfiguration::errorMinInterval = 0

minimum amount of iterations before the next refinement is allowed to happen in case of error based refinement

Referenced by sgpp::datadriven::RefinementMonitorFactory::createRefinementMonitor(), and sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig().

◆ levelPenalize

bool sgpp::base::AdaptivityConfiguration::levelPenalize = false

determines if finer grid levels should be penalized when finding points to refine

Referenced by sgpp::datadriven::ModelFittingClassification::fit(), and sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig().

◆ maxLevelType_

bool sgpp::base::AdaptivityConfiguration::maxLevelType_

refinement type: false: classic, true: maxLevel

Referenced by sgpp::datadriven::LearnerScenario::getAdaptivityConfiguration(), sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig(), hpx_main(), main(), sgpp::datadriven::LearnerScenario::setAdaptivityConfiguration(), sgpp::datadriven::FitterConfigurationLeastSquares::setupDefaults(), sgpp::datadriven::LearnerBaseSP::train(), and sgpp::datadriven::LearnerBase::train().

◆ noPoints_

size_t sgpp::base::AdaptivityConfiguration::noPoints_

max. number of points to be refined

Referenced by doAllRefinements(), sgpp::datadriven::LearnerSGDEOnOffParallel::doRefinementForAll(), sgpp::datadriven::ModelFittingClassification::fit(), sgpp::datadriven::LearnerScenario::getAdaptivityConfiguration(), sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig(), sgpp::datadriven::RegressionLearner::getMSE(), sgpp::datadriven::RefinementHandler::handleSurplusBasedRefinement(), hpx_main(), sgpp::datadriven::LearnerSGDEConfiguration::initConfig(), sgpp::datadriven::SparseGridDensityEstimatorConfiguration::initConfig(), sgpp::datadriven::LearnerSGDEConfiguration::LearnerSGDEConfiguration(), main(), sgpp::datadriven::LearnerScenario::setAdaptivityConfiguration(), sgpp::datadriven::FitterConfigurationLeastSquares::setupDefaults(), sgpp::datadriven::SparseGridDensityEstimatorConfiguration::SparseGridDensityEstimatorConfiguration(), sgpp::datadriven::LearnerSGD::train(), sgpp::datadriven::LearnerSVM::train(), sgpp::datadriven::LearnerBaseSP::train(), sgpp::datadriven::LearnerBase::train(), sgpp::datadriven::LearnerSGDE::train(), sgpp::datadriven::SparseGridDensityEstimator::train(), and sgpp::datadriven::LearnerSGDE::trainOnline().

◆ numRefinements_

size_t sgpp::base::AdaptivityConfiguration::numRefinements_

number of refinements

Referenced by sgpp::datadriven::RefinementHandler::checkRefinementNecessary(), sgpp::datadriven::RosenblattTransformation::createSGDELearner(), doAllRefinements(), sgpp::datadriven::LearnerScenario::getAdaptivityConfiguration(), sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig(), hpx_main(), sgpp::datadriven::LearnerSGDEConfiguration::initConfig(), sgpp::datadriven::SparseGridDensityEstimatorConfiguration::initConfig(), sgpp::datadriven::LearnerSGDEConfiguration::LearnerSGDEConfiguration(), main(), sgpp::datadriven::ModelFittingClassification::refine(), sgpp::datadriven::LearnerScenario::setAdaptivityConfiguration(), sgpp::datadriven::FitterConfigurationDensityEstimation::setupDefaults(), sgpp::datadriven::FitterConfigurationLeastSquares::setupDefaults(), sgpp::datadriven::SparseGridDensityEstimatorConfiguration::SparseGridDensityEstimatorConfiguration(), sgpp::datadriven::LearnerSGD::train(), sgpp::datadriven::LearnerBaseSP::train(), sgpp::datadriven::LearnerBase::train(), sgpp::datadriven::RegressionLearner::train(), sgpp::datadriven::LearnerSGDE::train(), sgpp::datadriven::SparseGridDensityEstimator::train(), and sgpp::datadriven::LearnerSGDE::trainOnline().

◆ percent_

double sgpp::base::AdaptivityConfiguration::percent_ = 1.0

max. percent of points to be refined

Referenced by sgpp::datadriven::LearnerScenario::getAdaptivityConfiguration(), sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig(), hpx_main(), main(), sgpp::datadriven::LearnerScenario::setAdaptivityConfiguration(), sgpp::datadriven::FitterConfigurationLeastSquares::setupDefaults(), sgpp::datadriven::LearnerBaseSP::train(), and sgpp::datadriven::LearnerBase::train().

◆ precomputeEvaluations

bool sgpp::base::AdaptivityConfiguration::precomputeEvaluations = true

in case of zero corssing based refinement: determines if evaluations should be precomupted

Referenced by sgpp::datadriven::ModelFittingClassification::fit(), sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig(), and sgpp::datadriven::ModelFittingClassification::refine().

◆ refinementFunctorType

RefinementFunctorType sgpp::base::AdaptivityConfiguration::refinementFunctorType = RefinementFunctorType::Surplus

refinement indicator

Referenced by sgpp::datadriven::ModelFittingClassification::fit(), sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig(), sgpp::datadriven::ModelFittingDensityEstimation::getRefinementFunctor(), and sgpp::datadriven::ModelFittingClassification::refine().

◆ refinementPeriod

size_t sgpp::base::AdaptivityConfiguration::refinementPeriod = 1

refinement will be triggered each refinementPeriod instances (approximately) in case of non error based refinement

Referenced by sgpp::datadriven::RefinementMonitorFactory::createRefinementMonitor(), sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig(), and sgpp::datadriven::FitterConfigurationDensityEstimation::setupDefaults().

◆ scalingCoefficients

std::vector<double> sgpp::base::AdaptivityConfiguration::scalingCoefficients = std::vector<double>()

in case of data based refinements: determines the scaling coefficients for each class

Referenced by sgpp::datadriven::ModelFittingClassification::fit(), and sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig().

◆ threshold_

double sgpp::base::AdaptivityConfiguration::threshold_

refinement threshold for surpluses

Referenced by doAllRefinements(), sgpp::datadriven::ModelFittingClassification::fit(), sgpp::datadriven::LearnerScenario::getAdaptivityConfiguration(), sgpp::datadriven::DataMiningConfigParser::getFitterAdaptivityConfig(), sgpp::datadriven::RegressionLearner::getMSE(), hpx_main(), main(), sgpp::datadriven::LearnerScenario::setAdaptivityConfiguration(), sgpp::datadriven::FitterConfigurationLeastSquares::setupDefaults(), sgpp::datadriven::LearnerSGD::train(), sgpp::datadriven::LearnerSVM::train(), sgpp::datadriven::LearnerBaseSP::train(), sgpp::datadriven::LearnerBase::train(), sgpp::datadriven::LearnerSGDE::train(), sgpp::datadriven::SparseGridDensityEstimator::train(), and sgpp::datadriven::LearnerSGDE::trainOnline().

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