The LAPACK forum has moved to https://github.com/Reference-LAPACK/lapack/discussions.

Sparse and Dense Matrixes

Open discussion regarding features, bugs, issues, vendors, etc.

Sparse and Dense Matrixes

Postby Danesh_D » Wed Jul 18, 2007 8:52 am

Hi all,

I am wondering if ScaLAPACK has some additional types for Dense and Sparse matrices? I mean for example a large sparse matrix can be saved in a format to just eleminate zeros and save non-zero elements to save memory. Does ScaLAPACK support this?

Regards,

D.
Danesh_D
 
Posts: 31
Joined: Mon Jun 04, 2007 10:03 pm

Postby Julien Langou » Wed Jul 18, 2007 10:25 am

This is called sparse linear algebra and is not supported by LAPACK nor ScaLAPACK.
Libraries that supports sparse matrices are for example MUMPS for direct methods or
PESTc for iterative methods. They do save a lots of FLOPs and memory compared to
dense solvers.
For more libraries go at
http://www.netlib.org/utk/people/JackDongarra/la-sw.html
Julien.
Julien Langou
 
Posts: 835
Joined: Thu Dec 09, 2004 12:32 pm
Location: Denver, CO, USA

Postby Danesh_D » Wed Jul 18, 2007 2:19 pm

Thank Julian.
I would also like to know,

1. Is it possible to use two libraries like GMM++ and ScaLAPACK at the same time? Will it be efficient to be used on a parallel (grid) computer sysem?

2. What is difference between "direct" and "iterative" solvers?

Regards,

Danesh
Danesh_D
 
Posts: 31
Joined: Mon Jun 04, 2007 10:03 pm

Postby semihozmen » Fri Mar 28, 2008 6:16 am

For direct solvers you should search for explicit solvers which are generally used for linear system of equations. On contrary for iterative ones the synonym is implicit which means a nonlinearity occurs..
semihozmen
 
Posts: 14
Joined: Mon Sep 10, 2007 4:58 am


Return to User Discussion

Who is online

Users browsing this forum: No registered users and 4 guests