sgpp::datadriven::OperationMakePositive Class Reference

This class enforces the function value range of a sparse grid function to be larger than 0. More...

#include <OperationMakePositive.hpp>

Public Types
typedef std::map< size_t, base::HashGridPoint >	gridPointCandidatesMap

Public Member Functions
std::vector< size_t > &	getAddedGridPoints ()
std::vector< size_t > &	getAddedGridPointsForPositivity ()
OperationMakePositiveCandidateSetAlgorithm &	getCandidateSetAlgorithm ()
void	initialize (base::Grid &grid, base::DataVector &alpha)
	initializes the operation More...
void	makeCurrentNodalValuesPositive (base::Grid &grid, base::DataVector &alpha, double tol=-1e-14)
	Enforce the function values at each grid point to larger than the specified tolerance. More...
void	makePositive (base::Grid &grid, base::DataVector &alpha, bool forcePositiveNodalValues=false)
	Make the sparse grid function defined by grid and coefficient vector positive. More...
size_t	numAddedGridPoints ()
size_t	numAddedGridPointsForPositivity ()
base::DataVector &	numAddedGridPointsForPositivityPerIteration ()
	OperationMakePositive (MakePositiveCandidateSearchAlgorithm candidateSearchAlgorithm=MakePositiveCandidateSearchAlgorithm::IntersectionsJoin, MakePositiveInterpolationAlgorithm interpolationAlgorithm=MakePositiveInterpolationAlgorithm::SetToZero, bool generateConsistentGrid=true, bool verbose=false, sgpp::optimization::ScalarFunction *f=nullptr)
	Constructor. More...
virtual	~OperationMakePositive ()
	Descrutor. More...

Detailed Description

This class enforces the function value range of a sparse grid function to be larger than 0.

It uses a discretization based approach where we add the minimum amount of full grid points to enforce the positivity.

Member Typedef Documentation

gridPointCandidatesMap

typedef std::map<size_t, base::HashGridPoint> sgpp::datadriven::OperationMakePositive::gridPointCandidatesMap

Constructor & Destructor Documentation

OperationMakePositive()

sgpp::datadriven::OperationMakePositive::OperationMakePositive ( MakePositiveCandidateSearchAlgorithm  candidateSearchAlgorithm = MakePositiveCandidateSearchAlgorithm::IntersectionsJoin, MakePositiveInterpolationAlgorithm  interpolationAlgorithm = MakePositiveInterpolationAlgorithm::SetToZero, bool  generateConsistentGrid = true, bool  verbose = false, sgpp::optimization::ScalarFunction *  f = nullptr  )

explicit

Constructor.

Parameters

candidateSearchAlgorithm	defines how to generate the full grid candidate set
interpolationAlgorithm	defines how to compute the coefficients for the new grid points
generateConsistentGrid	define if the hierarchical ancestors of all new grid points are inserted as well
verbose	print information or not
f	scalar function to be interpolated

~OperationMakePositive()

sgpp::datadriven::OperationMakePositive::~OperationMakePositive ( )

virtual

Descrutor.

Member Function Documentation

getAddedGridPoints()

std::vector< size_t > & sgpp::datadriven::OperationMakePositive::getAddedGridPoints

(

)

Returns

vector containing the indices of the added grid points

getAddedGridPointsForPositivity()

std::vector< size_t > & sgpp::datadriven::OperationMakePositive::getAddedGridPointsForPositivity

(

)

Returns

vector containing the indices which have just been added for positivity

getCandidateSetAlgorithm()

OperationMakePositiveCandidateSetAlgorithm & sgpp::datadriven::OperationMakePositive::getCandidateSetAlgorithm

(

)

Returns

candidate set algorithm

initialize()

void sgpp::datadriven::OperationMakePositive::initialize	(	base::Grid &	grid,
		base::DataVector &	alpha
	)

initializes the operation

Parameters

grid	Grid
alpha	coefficients

References sgpp::datadriven::FullGrid, sgpp::base::HashGridStorage::getDimension(), sgpp::base::HashGridStorage::getMaxLevel(), sgpp::base::Grid::getStorage(), python.utils.sg_projections::gridStorage, sgpp::datadriven::HybridFullIntersections, sgpp::datadriven::InterpolateBoundaries1d, sgpp::datadriven::InterpolateExp, sgpp::datadriven::InterpolateFunction, sgpp::datadriven::Intersections, sgpp::datadriven::IntersectionsJoin, and sgpp::datadriven::SetToZero.

Referenced by makePositive().

makeCurrentNodalValuesPositive()

void sgpp::datadriven::OperationMakePositive::makeCurrentNodalValuesPositive	(	base::Grid &	grid,
		base::DataVector &	alpha,
		double	tol = -1e-14
	)

Enforce the function values at each grid point to larger than the specified tolerance.

The ones which are not are set to zero. For this function we need the hierarchization and dechierarchization operations.

Parameters

grid	grid
alpha	coefficient vector
tol	tolerance for negativity

References alpha, sgpp::op_factory::createOperationEvalNaive(), sgpp::op_factory::createOperationHierarchisation(), sgpp::op_factory::createOperationMultipleEval(), python.statsfileInfo::data, sgpp::datadriven::DEFAULT, sgpp::base::HashGridStorage::getCoordinates(), sgpp::base::HashGridStorage::getDimension(), sgpp::base::HashGridStorage::getPoint(), sgpp::base::HashGridStorage::getSequenceNumber(), sgpp::base::HashGridStorage::getSize(), sgpp::base::DataVector::getSize(), sgpp::base::Grid::getStorage(), sgpp::base::Grid::getType(), grid(), python.utils.sg_projections::gridStorage, python.statsfileInfo::i, sgpp::base::HashGridStorage::insert(), sgpp::base::HashGridStorage::isContaining(), sgpp::base::Linear, numAddedGridPointsForPositivity(), sgpp::base::HashGridStorage::recalcLeafProperty(), sgpp::base::DataVector::resizeZero(), sgpp::base::DataMatrix::setRow(), and sgpp::datadriven::STREAMING.

Referenced by makePositive().

makePositive()

void sgpp::datadriven::OperationMakePositive::makePositive	(	base::Grid &	grid,
		base::DataVector &	alpha,
		bool	forcePositiveNodalValues = false
	)

Make the sparse grid function defined by grid and coefficient vector positive.

Parameters

grid	Grid where the new grid is stored
alpha	coefficient vector of new grid
forcePositiveNodalValues	nodal values are forced to be non-negative or not

References sgpp::base::Grid::getSize(), sgpp::base::Grid::getType(), initialize(), sgpp::base::Linear, sgpp::base::LinearBoundary, sgpp::base::LinearClenshawCurtis, sgpp::base::LinearClenshawCurtisBoundary, sgpp::base::LinearL0Boundary, sgpp::base::LinearTruncatedBoundary, makeCurrentNodalValuesPositive(), numAddedGridPoints(), and numAddedGridPointsForPositivity().

Referenced by sgpp::datadriven::SparseGridDensityEstimator::initialize().

numAddedGridPoints()

size_t sgpp::datadriven::OperationMakePositive::numAddedGridPoints

(

)

Returns

number of newly added grid points

Referenced by makePositive().

numAddedGridPointsForPositivity()

size_t sgpp::datadriven::OperationMakePositive::numAddedGridPointsForPositivity

(

)

Returns

number of newly added grid points for guaranteeing positivity

Referenced by makeCurrentNodalValuesPositive(), and makePositive().

numAddedGridPointsForPositivityPerIteration()

base::DataVector & sgpp::datadriven::OperationMakePositive::numAddedGridPointsForPositivityPerIteration

(

)

Returns

number of newly added grid points for guaranteeing positivity per iteration

---

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

• datadriven/src/sgpp/datadriven/operation/hash/simple/OperationMakePositive.hpp
• datadriven/src/sgpp/datadriven/operation/hash/simple/OperationMakePositive.cpp