[SciPy-user] CSparse: software for upcoming book on sparse direct methods

Nils Wagner nwagner at mecha.uni-stuttgart.de
Mon Mar 6 01:44:54 CST 2006


This might be of interest.

 Nils

From: Tim Davis <davis at cise.ufl.edu>
Date: Tue, 28 Feb 2006 13:46:36 -0500
Subject: CSparse: software for upcoming book on sparse direct methods

I would like to announce the release of a sparse matrix package,
CSparse, that I've written for my upcoming book, "Direct Methods
for Sparse Linear Systems," to appear in SIAM's Series on the
Fundamentals of Algorithms sometime this year.

It demonstrates a wide range of sparse matrix algorithms in as
concise a code as possible (sparse LU with partial pivoting is
only 165 lines, for example, not including the fill-reducing
ordering).  The algorithms are asymptotically optimal (in most
cases), and have comparable performance with existing methods
(with exception of the sparse LU, Cholesky, and QR factorizations).

One section of the book provides an overview of a wide range of
available software for sparse direct methods, with links to
high-performance sparse LU, Cholesky, and QR factorization codes:
http://www.cise.ufl.edu/research/sparse/codes

A MATLAB interface is included (including a pretty color "spy",
and an easy MATLAB interface to the UF Sparse Matrix Collection).

See http://www.cise.ufl.edu/research/sparse/CSparse or
http://www.cise.ufl.edu/research/sparse/CXSparse for an extended
version (for complex matrices).

Tim Davis
University of Florida



More information about the SciPy-user mailing list