scilin

Scilab Scilin Toolbox

Purpose

Scilin is a Scilab toolbox for sparse linear systems. It provides preconditionners and iterative methods for the sparse linear equations.

Scilin contains:

• Incomplete LU factorisations
• Iterative solvers
• Matrix Market I/O functions
• Miscellaneous

This project is OBSOLETE.

• imsls : Iterative Methods for Sparse Linear Equations http://forge.scilab.org/index.php/p/imsls/
• matmark : Matrix Market http://forge.scilab.org/index.php/p/matmark/
• spilu : Sparse Incomplete LU Preconditionners http://forge.scilab.org/index.php/p/spilu/

Features

• Iterative solvers

• bicgstab: BIConjugate Gradient STABilized method
• cgs: Conjugate Gradient Squared method
• cheby: CHEBYshev method
• gmres: Generalized Minimal RESidual method
• jacobi: JACOBI method
• qmr: Quasi Minimal Residual method
• sor: Successive Over-Relaxation method
• Incomplete LU factorisation

• ilut: Incomplete LU factorization with dual Truncation strategy
• ilutp: ilut with column Pivoting
• ilud: ilu with single dropping + Diagonal compensation ( ~milut )
• iludp: ilud with column Pivoting
• iluk: level-k ilu
• ilu0: simple ilu(0) preconditioning
• milu0: milu(0) preconditioning
• Matrix Market I/O functions

• mminfo: extracts storage information from a Matrix Market file (MM file)
• mmread: reads contents of MM file into sparse or dense format
• mmwrite: writes sparse or dense matrix to a MM file
• Miscellaneous functions

• det: returns the determinant of a sparse matrix
• nnz: gives the number of non-nul elements of a sparse matrix
• Other functions

• gen57pt : generates 5-point and 7-point matrices discretizing elliptic operators
• splsolve: ?
• spusolve: ?
• triangular: see if a sparse matrix is upper or lower triangular

Contents

• precond/: This directory contains interfaces of robust preconditioner routines using dual thresholding for dropping strategy. The routines come from the tool-box SPARSKIT.

• solvers/: This directory contains several basic iterative linear system solvers, written in Scilab language. The routines are adaptations of the mltemplates library.

• matio/: This directory contains some matrix generation routines from SPARSKIT and some matrix market input/output routines, adapted in Scilab language.

• addon/: This directory contains some useful functions, such as the determinant or the number of non null elements of a sparse matrix. The function %sp_l_s.sci allow now a upper and lower triangular resolution with optimized algorithms.

• demos/: This directory contains four demos using solvers and preconditioners with some typical matrices. ( see the directory /demos/matrices )

History

This module has been developed by Aladin Group (IRISA-INRIA) from a basic tool-kit for sparse matrix computations, SPARSKIT (developed by Y.Saad, University of Minnesota) and from the mltemplates library

Licence

This toolbox is released under the terms of the CeCILL license : http://www.cecill.info/licences/Licence_CeCILL_V2-en.txt

Authors

• 1993 - Univ. of Tennessee and Oak Ridge National Laboratory
• 2000 - 2001 - INRIA - Aladin Group
• 2010 - DIGITEO - Michael Baudin

Bibliography

• http://www.netlib.org/templates/
• http://graal.ens-lyon.fr/~jylexcel/scilab-sparse/meeting07/