PARDISOFactorization Class Reference

Factorization with DirectType::PARDISO. More...

#include <pardiso_solve.hh>

Detailed Description

Factorization with DirectType::PARDISO.

Important: can only be used with an appropriate license file! Set PARDISO_LIC_PATH to the directory where it is located.

Definition at line 25 of file fem/pardiso_solve.hh.

Public Types
enum	MatrixStructure { STRUCTURALLY_SYMMETRIC = 1 , SYMMETRIC_POSITIVE_DEFINITE = 2 , SYMMETRIC_INDEFINITE = -2 , NONSYMMETRIC = 11 }

Public Member Functions
	PARDISOFactorization (int n, int mtype_, std::vector< int > const &ridx, std::vector< int > const &cidx, std::vector< double > const &values)
	Version of constructor leaving input data in triplet format (aka coordinate format) unchanged. More...
	~PARDISOFactorization ()
void	solve (std::vector< double > const &b, std::vector< double > &x) const

Member Enumeration Documentation

MatrixStructure

enum PARDISOFactorization::MatrixStructure

Matrix structures accepted by the solver. Note that for symmetric matrices (SYMMETRIC_POSITIVE_DEFINITE, SYMMETRIC_INDEFINITE) the upper triangular part has to be given.

Enumerator
STRUCTURALLY_SYMMETRIC
SYMMETRIC_POSITIVE_DEFINITE
SYMMETRIC_INDEFINITE
NONSYMMETRIC

Definition at line 34 of file fem/pardiso_solve.hh.

Constructor & Destructor Documentation

PARDISOFactorization()

PARDISOFactorization::PARDISOFactorization	(	int	n,
		int	mtype_,
		std::vector< int > const &	ridx,
		std::vector< int > const &	cidx,
		std::vector< double > const &	values
	)

Version of constructor leaving input data in triplet format (aka coordinate format) unchanged.

Construction is factorization!

Parameters

n	size of the (square) matrix, i.e. the number of rows
mtype_	Matrix Type: several options available: 1 real structurally symmetric 2 real symmetric positive definite -2 real symmetric indefinite 11 real nonsymmetric 3 complex structurally symmetric 4 complex symmetric positive definite -4 complex symmetric indefinite 13 complex nonsymmetric
ridx	row indices
cidx	column indices
values	entry values

~PARDISOFactorization()

PARDISOFactorization::~PARDISOFactorization

(

)

Member Function Documentation

solve()

void PARDISOFactorization::solve	(	std::vector< double > const &	b,
		std::vector< double > &	x
	)		const

Solves the system for the given right hand side

• b.
• x is resized to the number of matrix columns.

Referenced by Kaskade::HigherOrderRecovery< Grid, Space >::getErrorFunction().

---

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

• fem/pardiso_solve.hh