QR factorization. More...

#include <sparseqr.hh>

Collaboration diagram for sparseqr::QR:

Public Member Functions
void	BackSolve (int rank, double *x)
	Performes backsolve.
int	givens (int, int, int)
	Performs Given's rotations.
	QR (Packed qrpmatrix, int qrnr, int qrnc, double qrb=NULL, Qmatrix qrq=NULL, int qrprow=NULL, int *qrpcol=NULL, double qrztol=DBL_EPSILON, double qrrtol=0.0)
	Constructor.
void	RemoveEntry (int, int)
Smatrix *	RemoveEntry (int, int, Smatrix , Smatrix )
int	RemoveMarkedEntries (int)
int	sparseqr ()
	Performs QR factorization.
	~QR ()
Public Attributes
double *	b
bool	del_pcol
bool	del_prow
	Stores if `delete`[] of `prow` and `pcol` is allowed.
long int	fillin
long int	flops
int	nc
long int	ngivens
int	nr
unsigned *	nrow
int *	pcol
Packed *	pmatrix
int *	prow
Qmatrix *	q
double	rtol
Smatrix **	scol
Smatrix **	srow
double	ztol

Detailed Description

QR factorization.

See also:: Homepage of SparseQR
Thomas H. Robey and Deborah L. Sulsky, Row Ordering for Sparse QR Decomposition, SIAM J. Matrix Anal. and Appl., Vol. 15, No. 4, pp. 1208-1225, October 1994.

Definition at line 122 of file sparseqr.hh.

Constructor & Destructor Documentation

Constructor.

Parameters:

qrpmatrix	Matrix which should be factorized in sparse format
qrnr	Number of rows of the matrix
qrnc	Number of columns of the matrix
qrb	Right hand side
qrq	Space to store the Given's rotations. Does not need to be given, if they should not be stored (ie. solving for just 1 right hand side).
qrprow	Permutation vector for row indices. If not given, is allocated by the class.
qrpcol	Permutation vector for column indices. If not given, is allocated by the class.
qrztol	Tolerance for use in determining zero entries when using Given's rotations. If zero or negative, only zero entries are deleted.
rtol	Tolerance for use in determining rank tolerance by testing the diagonal entry in the final R matrix.