Polynomial preconditioner implementation. More...

#include <polynomial_precond.hpp>

Inheritance diagram for FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >:

Public Types
typedef SolverBase< VectorType >	BaseClass
	Our base class. More...

typedef MatrixType::DataType	DataType
	The floating point precision. More...

typedef Filter_	FilterType
	The filter type. More...

typedef Matrix_	MatrixType
	The matrix type. More...

typedef MatrixType::VectorTypeL	VectorType
	The type of vector this solver can be applied to. More...

Public Member Functions
	PolynomialPrecond (const MatrixType &matrix, const FilterType &filter, Index m, DataType omega=DataType(1))
	Constructor. More...

	PolynomialPrecond (const String &section_name, const PropertyMap *section, const MatrixType &matrix, const FilterType &filter)
	Constructor using a PropertyMap. More...

virtual	~PolynomialPrecond ()
	Empty virtual destructor. More...

virtual Status	apply (Matrix_::VectorTypeL &vec_cor, const Matrix_::VectorTypeL &vec_def)=0
	Solver application method. More...

virtual Status	apply (VectorType &vec_cor, const VectorType &vec_def) override
	Solver application method. More...

virtual void	done ()
	Finalization method. More...

virtual void	done_numeric ()
	Numeric finalization method. More...

virtual void	done_symbolic () override
	Symbolic finalization method. More...

virtual void	init ()
	Initialization method. More...

virtual void	init_numeric () override
	Numeric initialization method. More...

virtual void	init_symbolic () override
	Symbolic initialization method. More...

virtual String	name () const override
	Returns a descriptive string. More...

void	set_m (Index m)
	Sets the polynomial order parameter. More...

void	set_omega (DataType omega)
	Sets the damping parameter. More...

Protected Attributes
VectorType	_aux1
	auxiliary vectors More...

VectorType	_aux2

VectorType	_aux3

const FilterType &	_filter
	The filter for projecting solution and defect to subspaces. More...

VectorType	_inv_diag
	The component-wise inverted diagonal of _matrix. More...

Index	_m
	order m of preconditioner More...

const MatrixType &	_matrix
	The system matrix. More...

DataType	_omega
	The damping parameter for the internal jacobi preconditioner. More...

Detailed Description

template<typename Matrix_, typename Filter_>
class FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >

Polynomial preconditioner implementation.

This class represents the Neumann-Polynomial-Preconditioner \(M^{-1} = \sum_{k=0}^m (I - \tilde M^{-1}A)^k \tilde M^{-1}\)

Note: As of now, M is a hardcoded Jacobi (main diagonal inverse)

Definition at line 25 of file polynomial_precond.hpp.

Member Typedef Documentation

◆ BaseClass

template<typename Matrix_ , typename Filter_ >

typedef SolverBase<VectorType> FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::BaseClass

Our base class.

Definition at line 38 of file polynomial_precond.hpp.

◆ DataType

template<typename Matrix_ , typename Filter_ >

typedef MatrixType::DataType FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::DataType

The floating point precision.

Definition at line 36 of file polynomial_precond.hpp.

◆ FilterType

template<typename Matrix_ , typename Filter_ >

typedef Filter_ FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::FilterType

The filter type.

Definition at line 32 of file polynomial_precond.hpp.

◆ MatrixType

template<typename Matrix_ , typename Filter_ >

typedef Matrix_ FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::MatrixType

The matrix type.

Definition at line 30 of file polynomial_precond.hpp.

◆ VectorType

template<typename Matrix_ , typename Filter_ >

typedef MatrixType::VectorTypeL FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::VectorType

The type of vector this solver can be applied to.

Definition at line 34 of file polynomial_precond.hpp.

Constructor & Destructor Documentation

◆ PolynomialPrecond() [1/2]

template<typename Matrix_ , typename Filter_ >

FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::PolynomialPrecond	(	const MatrixType &	matrix,
		const FilterType &	filter,
		Index	m,
		DataType	omega = `DataType(1)`
	)

inlineexplicit

Constructor.

Parameters

[in]	matrix	The source matrix.
[in]	filter	The system filter.
[in]	m	The order of the polynom
[in]	omega	The damping parameter for the internal jacobi preconditioner.

Definition at line 70 of file polynomial_precond.hpp.

◆ PolynomialPrecond() [2/2]

template<typename Matrix_ , typename Filter_ >

FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::PolynomialPrecond	(	const String &	section_name,
		const PropertyMap *	section,
		const MatrixType &	matrix,
		const FilterType &	filter
	)

inlineexplicit

Constructor using a PropertyMap.

Parameters

[in]	section_name	The name of the config section, which it does not know by itself
[in]	section	A pointer to the PropertyMap section configuring this solver
[in]	matrix	The system matrix.
[in]	filter	The system filter.

Returns: A shared pointer to a new PolynomialPrecond object.

Definition at line 96 of file polynomial_precond.hpp.

References FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::name(), FEAT::PropertyMap::query(), FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::set_m(), FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::set_omega(), and XABORTM.

◆ ~PolynomialPrecond()

template<typename Matrix_ , typename Filter_ >

virtual FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::~PolynomialPrecond ( )

inlinevirtual

Empty virtual destructor.

Definition at line 128 of file polynomial_precond.hpp.

Member Function Documentation

◆ apply() [1/2]

virtual Status FEAT::Solver::SolverBase< Matrix_::VectorTypeL >::apply	(	Matrix_::VectorTypeL &	vec_cor,
		const Matrix_::VectorTypeL &	vec_def
	)

pure virtualinherited

Solver application method.

This method applies the solver represented by this object onto a given defect vector and returns the corresponding correction vector.

Note: Solvers which derive from the IterativeSolver base class also provide a correct() method which corrects an initial solution instead of starting with the null vector.

Parameters

[out]	vec_cor	The vector that shall receive the solution of the linear system. It is assumed to be allocated, but its numerical contents may be undefined upon calling this method.
[in]	vec_def	The vector that represents the right-hand-side of the linear system to be solved.

Attention: vec_cor and vec_def must not refer to the same vector object!

Returns: A Status code that represents the status of the solution step.

Implemented in FEAT::Solver::AmaVanka< Matrix_, Filter_ >.

◆ apply() [2/2]

template<typename Matrix_ , typename Filter_ >

virtual Status FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::apply	(	VectorType &	vec_cor,
		const VectorType &	vec_def
	)

inlineoverridevirtual

Solver application method.

This method applies the solver represented by this object onto a given defect vector and returns the corresponding correction vector.

Note: Solvers which derive from the IterativeSolver base class also provide a correct() method which corrects an initial solution instead of starting with the null vector.

Parameters

[out]	vec_cor	The vector that shall receive the solution of the linear system. It is assumed to be allocated, but its numerical contents may be undefined upon calling this method.
[in]	vec_def	The vector that represents the right-hand-side of the linear system to be solved.

Attention: vec_cor and vec_def must not refer to the same vector object!

Returns: A Status code that represents the status of the solution step.

Definition at line 193 of file polynomial_precond.hpp.

References FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::_aux1, FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::_filter, FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::_inv_diag, FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::_m, FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::_matrix, and FEAT::Solver::success.

◆ done()

virtual void FEAT::Solver::SolverBase< Matrix_::VectorTypeL >::done ( )

inlinevirtualinherited

Finalization method.

This function performs both the symbolic and numeric finalization, i.e. it simply performs

this->done_numeric();
this->done_symbolic();

Definition at line 283 of file base.hpp.

◆ done_numeric()

virtual void FEAT::Solver::SolverBase< Matrix_::VectorTypeL >::done_numeric ( )

inlinevirtualinherited

Numeric finalization method.

This method is called to release any data allocated in the numeric initialization step.

Reimplemented in FEAT::Solver::GenericUmfpack< Matrix_ >.

Definition at line 246 of file base.hpp.

◆ done_symbolic()

template<typename Matrix_ , typename Filter_ >

virtual void FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::done_symbolic ( )

inlineoverridevirtual

Symbolic finalization method.

This method is called to release any data allocated in the symbolic initialization step.

Reimplemented from FEAT::Solver::SolverBase< Matrix_::VectorTypeL >.

Definition at line 148 of file polynomial_precond.hpp.

References FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::_aux1, and FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::_inv_diag.

◆ init()

virtual void FEAT::Solver::SolverBase< Matrix_::VectorTypeL >::init ( )

inlinevirtualinherited

Initialization method.

This function performs both the symbolic and numeric initialization, i.e. it simply performs

this->init_symbolic();
this->init_numeric();

Definition at line 268 of file base.hpp.

◆ init_numeric()

template<typename Matrix_ , typename Filter_ >

virtual void FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::init_numeric ( )

inlineoverridevirtual

Numeric initialization method.

This method is called to perform numeric initialization of the solver.
Before this function can be called, the symbolic initialization must be performed.

Reimplemented from FEAT::Solver::SolverBase< Matrix_::VectorTypeL >.

Definition at line 157 of file polynomial_precond.hpp.

References FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::_inv_diag, FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::_matrix, and FEAT::Solver::PolynomialPrecond< Matrix_, Filter_ >::_omega.