toeplitz_nofft.c File Reference

Contains basic product without using ffts for Toeplitz algebra. More...

Go to the source code of this file.

Functions
int	stmm_simple_basic (double **V, int n, int m, double *T, int lambda, double **TV)
	Perform the product of a Toeplitz matrix by a matrix without using FFT's. More...
int	stmm_simple_core (double **V, int n, int m, double *T, int blocksize, int lambda, int nfft, int flag_offset)

Variables
int	PRINT_RANK

Detailed Description

Contains basic product without using ffts for Toeplitz algebra.

version 1.1b, July 2012

Author

Frederic Dauvergne

Project: Midapack library, ANR MIDAS'09 - Toeplitz Algebra module Purpose: Provide Toeplitz algebra tools suitable for Cosmic Microwave Background (CMB) data analysis.

Note

Copyright (c) 2010-2012 APC CNRS Université Paris Diderot

This program is free software; you can redistribute it and/or modify it under the terms

of the GNU Lesser General Public License as published by the Free Software Foundation;

either version 3 of the License, or (at your option) any later version. This program is

distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even

the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU

Lesser General Public License for more details.

You should have received a copy of the GNU Lesser General Public License along with this

program; if not, see http://www.gnu.org/licenses/lgpl.html

For more information about ANR MIDAS'09 project see :

http://www.apc.univ-paris7.fr/APC_CS/Recherche/Adamis/MIDAS09/index.html

ACKNOWLEDGMENT: This work has been supported in part by the French National Research

Agency (ANR) through COSINUS program (project MIDAS no. ANR-09-COSI-009).

Log: toeplitz*.c

Revision 1.0b 2012/05/07 Frederic Dauvergne (APC) Official release 1.0beta. The first installement of the library is the Toeplitz algebra module.

Revision 1.1b 2012/07/- Frederic Dauvergne (APC)

• mpi_stbmm allows now rowi-wise order per process datas and no-blocking communications.
• OMP improvment for optimal cpu time.
• bug fixed for OMP in the stmm_basic routine.
• distcorrmin is used to communicate only lambda-1 datas when it is needed.
• new reshaping routines using transformation functions in stmm. Thus, only one copy at most is needed.
• tpltz_init improvement using define_nfft and define_blocksize routines.
• add Block struture to define each Toeplitz block.
• add Flag structure and preprocessing parameters to define the computational strategy. All the flag parameters are then available directly from the API.

Revision 1.2b 2012/11/30 Frederic Dauvergne (APC)

• extend the mpi product routine to rowwise order data distribution. This is now allowing tree kinds of distribution.
• add int64 for some variables to extend the global volume of data you can use.
• Openmp improvments.
• Add toeplitz_wizard.c, which contains a set of easy to use routines with defined structures.

Definition in file toeplitz_nofft.c.

Function Documentation

stmm_simple_basic()

int stmm_simple_basic	(	double **	V,
		int	n,
		int	m,
		double *	T,
		int	lambda,
		double **	TV
	)

Perform the product of a Toeplitz matrix by a matrix without using FFT's.

This routine multiplies the values directly between them. This exploit the fact that the bandwith is small compared to the matrix size. The number of operation is then no more than (lambda*2-1) multiplications and (lambda*2-1)-1 additions per row.

Definition at line 104 of file toeplitz_nofft.c.

stmm_simple_core()

int stmm_simple_core	(	double **	V,
		int	n,
		int	m,
		double *	T,
		int	blocksize,
		int	lambda,
		int	nfft,
		int	flag_offset
	)

Perform the stand alone product of a Toeplitz matrix by a matrix using the sliding window algorithm.

The product is performed block-by-block with a defined block size or a computed optimized blocksize. This routine is not used by th API.

Parameters

V	[input] data matrix (with the convention V(i,j)=V[i+j*n]) ; [out] result of the product TV
n	number of rows of V
m	number of columns of V
T	Toeplitz matrix data composed of the non-zero entries of his first row
blocksize	block size used in the sliding window algorithm
lambda	Toeplitz band width
nfft	number of simultaneous FFTs
flag_offset	flag to avoid extra 2*lambda padding to zeros on the edges

Definition at line 156 of file toeplitz_nofft.c.

Variable Documentation

PRINT_RANK

int PRINT_RANK

extern

Definition at line 117 of file toeplitz.c.