sgpp::optimization::sle_solver::Auto Class Reference

Automatic choice of external linear solver. More...

#include <Auto.hpp>

Inheritance diagram for sgpp::optimization::sle_solver::Auto:

Public Member Functions
bool	solve (SLE &system, base::DataVector &b, base::DataVector &x) const override
bool	solve (SLE &system, base::DataMatrix &B, base::DataMatrix &X) const override
	~Auto () override
	Destructor. More...
Public Member Functions inherited from sgpp::optimization::sle_solver::SLESolver
	SLESolver ()
	Constructor. More...
virtual	~SLESolver ()
	Destructor. More...

Static Public Attributes
static constexpr double	ESTIMATE_NNZ_ROWS_SAMPLE_SIZE = 0.05
	ratio of the rows (e.g. 0.1 = 10%) to use for sparsity estimation More...
static const size_t	MAX_DIM_FOR_FULL = 30000
	maximal matrix dimension to allow use of full solvers More...
static const size_t	MAX_DIM_FOR_GAUSSIAN = 200
	maximal matrix dimension to prefer GaussianElimination over BiCGStab More...
static constexpr double	MAX_NNZ_RATIO_FOR_SPARSE = 0.1
	maximal ratio of non-zero entries for sparse solvers More...

Detailed Description

Automatic choice of external linear solver.

Constructor & Destructor Documentation

~Auto()

sgpp::optimization::sle_solver::Auto::~Auto ( )

override

Destructor.

Member Function Documentation

solve() [1/2]

bool sgpp::optimization::sle_solver::Auto::solve ( SLE &  system, base::DataVector &  b, base::DataVector &  x  ) const

overridevirtual

Parameters

	system	system to be solved
	b	right-hand side
[out]	x	solution to the system

Returns

whether all went well (false if errors occurred)

Implements sgpp::optimization::sle_solver::SLESolver.

References sgpp::base::DataVector::getPointer(), sgpp::base::DataVector::getSize(), and friedman::X.

Referenced by BSplineCoefficientGridFunction(), BSplineTensorCoefficientGridFunction(), sgpp::combigrid::calculateInterpolationCoefficientsForConvertedExpUniformBoundaryCombigird(), sgpp::optimization::OperationMultipleHierarchisationLinear::doHierarchisation(), sgpp::optimization::OperationMultipleHierarchisationModLinear::doHierarchisation(), sgpp::optimization::OperationMultipleHierarchisationWaveletBoundary::doHierarchisation(), sgpp::optimization::OperationMultipleHierarchisationModBspline::doHierarchisation(), sgpp::optimization::OperationMultipleHierarchisationLinearBoundary::doHierarchisation(), sgpp::optimization::OperationMultipleHierarchisationBspline::doHierarchisation(), sgpp::optimization::OperationMultipleHierarchisationWavelet::doHierarchisation(), sgpp::optimization::OperationMultipleHierarchisationModWavelet::doHierarchisation(), sgpp::optimization::OperationMultipleHierarchisationBsplineBoundary::doHierarchisation(), sgpp::optimization::OperationMultipleHierarchisationLinearClenshawCurtis::doHierarchisation(), sgpp::optimization::OperationMultipleHierarchisationModBsplineClenshawCurtis::doHierarchisation(), sgpp::optimization::OperationMultipleHierarchisationBsplineClenshawCurtis::doHierarchisation(), and sgpp::combigrid::FullGridPCESummationStrategy< V >::eval().

solve() [2/2]

bool sgpp::optimization::sle_solver::Auto::solve ( SLE &  system, base::DataMatrix &  B, base::DataMatrix &  X  ) const

overridevirtual

Parameters

	system	system to be solved
	B	matrix of right-hand sides
[out]	X	matrix of solutions to the systems

Returns

whether all went well (false if errors occurred)

Reimplemented from sgpp::optimization::sle_solver::SLESolver.

References sgpp::optimization::sle_solver::addSLESolver(), ESTIMATE_NNZ_ROWS_SAMPLE_SIZE, sgpp::optimization::SLE::getDimension(), sgpp::optimization::Printer::getInstance(), python.statsfileInfo::i, sgpp::optimization::SLE::isMatrixEntryNonZero(), python.utils.statsfile2gnuplot::j, MAX_DIM_FOR_FULL, MAX_DIM_FOR_GAUSSIAN, MAX_NNZ_RATIO_FOR_SPARSE, sgpp::optimization::Printer::printStatusBegin(), sgpp::optimization::Printer::printStatusEnd(), sgpp::optimization::Printer::printStatusNewLine(), and sgpp::optimization::Printer::printStatusUpdate().

Member Data Documentation

ESTIMATE_NNZ_ROWS_SAMPLE_SIZE

constexpr double sgpp::optimization::sle_solver::Auto::ESTIMATE_NNZ_ROWS_SAMPLE_SIZE = 0.05

static

ratio of the rows (e.g. 0.1 = 10%) to use for sparsity estimation

Referenced by solve().

MAX_DIM_FOR_FULL

const size_t sgpp::optimization::sle_solver::Auto::MAX_DIM_FOR_FULL = 30000

static

maximal matrix dimension to allow use of full solvers

Referenced by solve().

MAX_DIM_FOR_GAUSSIAN

const size_t sgpp::optimization::sle_solver::Auto::MAX_DIM_FOR_GAUSSIAN = 200

static

maximal matrix dimension to prefer GaussianElimination over BiCGStab

Referenced by solve().

MAX_NNZ_RATIO_FOR_SPARSE

constexpr double sgpp::optimization::sle_solver::Auto::MAX_NNZ_RATIO_FOR_SPARSE = 0.1

static

maximal ratio of non-zero entries for sparse solvers

Referenced by solve().

---

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

• optimization/src/sgpp/optimization/sle/solver/Auto.hpp
• optimization/src/sgpp/optimization/sle/solver/Auto.cpp