Sparse direct solver for symmetric and unsymmetric matrices. More...

#include <pardiso.hh>

Inheritance diagram for concepts::Pardiso:

Public Types
typedef Cmplxtype< Real >::type	c_type
	Real type of data type. More...

enum	MATRIX_TYPE { STRUCT_SYMM = 1, SPD = 2, SYMM_INDEF = -2, UNSYMM = 11 }
	Type of matrix to solve. More...

typedef Realtype< Real >::type	r_type
	Real type of data type. More...

typedef Real	type
	Type of data, e.g. matrix entries. More...

Public Member Functions
virtual const uint	dimX () const
	Returns the size of the image space of the operator (number of rows of the corresponding matrix) More...

virtual const uint	dimY () const
	Returns the size of the source space of the operator (number of columns of the corresponding matrix) More...

virtual void	operator() ()
	Application operator without argument. More...

virtual void	operator() (const Function< c_type > &fncY, Function< c_type > &fncX)
	Application operator for complex function `fncY`. More...

virtual void	operator() (const Function< r_type > &fncY, Function< Real > &fncX)
	Application operator for real function `fncY`. More...

virtual void	operator() (const Function< Real > &fncY, Function< Real > &fncX)

void	operator() (const Vector< Real > &fncY, Vector< Real > &fncX)

	Pardiso (const SparseMatrix< Real > &matrix, enum MATRIX_TYPE type)
	Constructor. More...

virtual void	show_messages ()

virtual const Space< Real > &	spaceX () const

virtual const Space< Real > &	spaceY () const

virtual	~Pardiso ()

Protected Member Functions
virtual std::ostream &	info (std::ostream &os) const

Protected Attributes
uint	dimX_
	Dimension of image space and the source space. More...

uint	dimY_

Private Member Functions
void	errorcheck_ (int error) const throw (FortranException)
	Checks error condition of a PARDISO call and outputs a error message. More...

int	solver_pardiso_factor_ ()
	Calls PARDISO to factor the matrix (symbolically and numerically) More...

int	solver_pardiso_solve_ (double , double )
	Calls PARDISO to do the forward and backward substiution. More...

Private Attributes
void *	A_
	PARDISO data: matrix in Compressed Sparse Row format. More...

Array< Real >	a_
	PARDISO data: entries of the matrix. More...

int	error

bool	factored_
	Stores if the factorization was already computed before. More...

Array< size_t >	handle
	PARDISO data: parameter. More...

Array< uint >	ia_
	PARDISO data: row and column indices of the matrix. More...

Array< int >	iparam
	PARDISO data: parameter. More...

Array< uint >	ja_

int	matrix_number

int	matrix_type

int	max_factorizations
	PARDISO data: parameter. More...

int	max_threads

int	msglvl

const uint	n_
	System size. More...

int	nb

uint	nnz_
	PARDISO data: number of non-zeros. More...

const Space< Real > &	spcX_
	Image space. More...

const Space< Real > &	spcY_
	Source space. More...

bool	symmetric_
	Symmetric system? More...

Detailed Description

Sparse direct solver for symmetric and unsymmetric matrices.

This class interfaces to PARDISO [1]. Use the environment variable OMP_NUM_THREADS to change the number of threas used (default is 1).

The package PARDISO is a thread-safe, high-performance, robust, memory efficient and easy to use software for solving large sparse symmetric and unsymmetric linear systems of equations on shared memory multiprocessors. The solver uses a combination of left- and right-looking Level-3 BLAS supernode techniques to exploit pipelining parallelism and memory hierarchies. In order to improve sequential and parallel sparse numerical factorization performance, the algorithms are based on a Level-3 BLAS update and pipelining parallelism is exploited with a combination of left- and right-looking supernode techniques.

See also: [1] Olaf Schenk, Klaus G�rtner, Wolfgang Fichtner, Andreas Stricker, PARDISO: a high-performance serial and parallel sparse linear Linear Solvers in semiconductor device simulation, Future Generation Computer System 18 (2001), 69-78.; [2] Pardiso Homepage

Author: Philipp Frauenfelder, 2003

Test:: test::PardisoTest

Examples: hpFEM2d.cc.

Definition at line 43 of file pardiso.hh.

Member Typedef Documentation

◆ c_type

typedef Cmplxtype<Real >::type concepts::Operator< Real >::c_type

inherited

Real type of data type.

Definition at line 49 of file compositions.hh.

◆ r_type

typedef Realtype<Real >::type concepts::Operator< Real >::r_type

inherited

Real type of data type.

Definition at line 47 of file compositions.hh.

◆ type

typedef Real concepts::Operator< Real >::type

inherited

Type of data, e.g. matrix entries.

Definition at line 45 of file compositions.hh.

Member Enumeration Documentation

◆ MATRIX_TYPE

enum concepts::Pardiso::MATRIX_TYPE

Type of matrix to solve.

Enumerator
STRUCT_SYMM
SPD
SYMM_INDEF
UNSYMM

Definition at line 46 of file pardiso.hh.

Constructor & Destructor Documentation

◆ Pardiso()

concepts::Pardiso::Pardiso	(	const SparseMatrix< Real > &	matrix,
		enum MATRIX_TYPE	type
	)

Constructor.

Parameters

matrix	Matrix
type	Type of the matrix

◆ ~Pardiso()

virtual concepts::Pardiso::~Pardiso ( )

virtual

Member Function Documentation

◆ dimX()

virtual const uint concepts::Operator< Real >::dimX ( ) const

inlinevirtualinherited

Returns the size of the image space of the operator (number of rows of the corresponding matrix)

Definition at line 93 of file compositions.hh.

◆ dimY()

virtual const uint concepts::Operator< Real >::dimY ( ) const

inlinevirtualinherited

Returns the size of the source space of the operator (number of columns of the corresponding matrix)

Definition at line 98 of file compositions.hh.

◆ errorcheck_()

void concepts::Pardiso::errorcheck_	(	int	error	)	const
throw	(	FortranException
	)

private

Checks error condition of a PARDISO call and outputs a error message.

◆ info()

virtual std::ostream& concepts::Pardiso::info ( std::ostream & os ) const

protectedvirtual

Reimplemented from concepts::Operator< Real >.

◆ operator()() [1/5]

virtual void concepts::Operator< Real >::operator() ( )

virtualinherited

Application operator without argument.

Reimplemented in concepts::VecOperator< Real >.

◆ operator()() [2/5]

virtual void concepts::Operator< Real >::operator()	(	const Function< c_type > &	fncY,
		Function< c_type > &	fncX
	)

virtualinherited

Application operator for complex function fncY.

Computes fncX = A(fncY) where A is this operator. fncX becomes complex.

In derived classes its enough to implement the operator() for complex Operator's. If a real counterpart is not implemented, the function fncY is splitted into real and imaginary part and the application operator for real functions is called for each. Then the result is combined.

If in a derived class the operator() for complex Operator's is not implemented, a exception is thrown from here.

Reimplemented in concepts::VecOperator< Real >.

◆ operator()() [3/5]

virtual void concepts::Operator< Real >::operator()	(	const Function< r_type > &	fncY,
		Function< F > &	fncX
	)

virtualinherited

Application operator for real function fncY.

Computes fncX = A(fncY) where A is this operator.

fncX becomes the type of the operator, for real data it becomes real, for complex data it becomes complex.

In derived classes its enough to implement the operator() for real Operator's. If a complex counterpart is not implemented, the function fncY is transformed to a complex function and then the application operator for complex functions is called.

If in a derived class the operator() for real Operator's is not implemented, a exception is thrown from here.

Reimplemented in concepts::VecOperator< Real >.

◆ operator()() [4/5]

virtual void concepts::Pardiso::operator()	(	const Function< Real > &	fncY,
		Function< Real > &	fncX
	)

virtual

◆ operator()() [5/5]

void concepts::Pardiso::operator()	(	const Vector< Real > &	fncY,
		Vector< Real > &	fncX
	)

◆ show_messages()

virtual void concepts::Operator< Real >::show_messages ( )

inlinevirtualinherited

Definition at line 100 of file compositions.hh.

◆ solver_pardiso_factor_()

int concepts::Pardiso::solver_pardiso_factor_ ( )

private

Calls PARDISO to factor the matrix (symbolically and numerically)

◆ solver_pardiso_solve_()

int concepts::Pardiso::solver_pardiso_solve_	(	double *	,
		double *
	)

private

Calls PARDISO to do the forward and backward substiution.

◆ spaceX()

virtual const Space<Real>& concepts::Pardiso::spaceX ( ) const

inlinevirtual

Definition at line 64 of file pardiso.hh.

◆ spaceY()

virtual const Space<Real>& concepts::Pardiso::spaceY ( ) const

inlinevirtual

Definition at line 65 of file pardiso.hh.

Member Data Documentation

◆ A_

void* concepts::Pardiso::A_

private

PARDISO data: matrix in Compressed Sparse Row format.

Definition at line 83 of file pardiso.hh.

◆ a_

Array<Real> concepts::Pardiso::a_

private

PARDISO data: entries of the matrix.

Definition at line 85 of file pardiso.hh.

◆ dimX_

uint concepts::Operator< Real >::dimX_

protectedinherited

Dimension of image space and the source space.

Definition at line 104 of file compositions.hh.

◆ dimY_

uint concepts::Operator< Real >::dimY_

protectedinherited

Definition at line 104 of file compositions.hh.

◆ error

int concepts::Pardiso::error

private

Definition at line 93 of file pardiso.hh.

◆ factored_

bool concepts::Pardiso::factored_

private

Stores if the factorization was already computed before.

Definition at line 78 of file pardiso.hh.

◆ handle

Array<size_t> concepts::Pardiso::handle

private

PARDISO data: parameter.

Definition at line 89 of file pardiso.hh.

◆ ia_

Array<uint> concepts::Pardiso::ia_

private

PARDISO data: row and column indices of the matrix.

Definition at line 87 of file pardiso.hh.

◆ iparam

Array<int> concepts::Pardiso::iparam

private

PARDISO data: parameter.

Definition at line 91 of file pardiso.hh.

◆ ja_

Array<uint> concepts::Pardiso::ja_

private

Definition at line 87 of file pardiso.hh.

◆ matrix_number

int concepts::Pardiso::matrix_number

private

Definition at line 93 of file pardiso.hh.

◆ matrix_type

int concepts::Pardiso::matrix_type

private

Definition at line 93 of file pardiso.hh.

◆ max_factorizations

int concepts::Pardiso::max_factorizations

private

PARDISO data: parameter.

Definition at line 93 of file pardiso.hh.

◆ max_threads

int concepts::Pardiso::max_threads

private

Definition at line 94 of file pardiso.hh.

◆ msglvl

int concepts::Pardiso::msglvl

private

Definition at line 93 of file pardiso.hh.

◆ n_

const uint concepts::Pardiso::n_

private

System size.

Definition at line 70 of file pardiso.hh.

◆ nb

int concepts::Pardiso::nb

private

Definition at line 94 of file pardiso.hh.

◆ nnz_

uint concepts::Pardiso::nnz_

private

PARDISO data: number of non-zeros.

Definition at line 81 of file pardiso.hh.

◆ spcX_

const Space<Real>& concepts::Pardiso::spcX_

private

Image space.

Definition at line 72 of file pardiso.hh.

◆ spcY_

const Space<Real>& concepts::Pardiso::spcY_

private

Source space.

Definition at line 74 of file pardiso.hh.

◆ symmetric_

bool concepts::Pardiso::symmetric_

private

Symmetric system?

Definition at line 76 of file pardiso.hh.

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

operator/pardiso.hh

concepts::Pardiso Class ReferenceLinear Solvers

Public Types

Public Member Functions

Protected Member Functions

Protected Attributes

Private Member Functions

Private Attributes

Detailed Description

Member Typedef Documentation

◆ c_type

◆ r_type

◆ type

Member Enumeration Documentation

◆ MATRIX_TYPE

Constructor & Destructor Documentation

◆ Pardiso()

◆ ~Pardiso()

Member Function Documentation

◆ dimX()

◆ dimY()

◆ errorcheck_()

◆ info()

◆ operator()() [1/5]

◆ operator()() [2/5]

◆ operator()() [3/5]

◆ operator()() [4/5]

◆ operator()() [5/5]

◆ show_messages()

◆ solver_pardiso_factor_()

◆ solver_pardiso_solve_()

◆ spaceX()

◆ spaceY()

Member Data Documentation

◆ A_

◆ a_

◆ dimX_

◆ dimY_

◆ error

◆ factored_

◆ handle

◆ ia_

◆ iparam

◆ ja_

◆ matrix_number

◆ matrix_type

◆ max_factorizations

◆ max_threads

◆ msglvl

◆ n_

◆ nb

◆ nnz_

◆ spcX_

◆ spcY_

◆ symmetric_

concepts::Pardiso Class Reference
Linear Solvers