For Programmers: Free Programming Magazines  


Home > Archive > Fortran > May 2006 > A better direct sparse solver for linear system with a positive definite matrix









You are viewing an archived Text-only version of the thread. To view this thread in it's original format and/or if you want to reply to this thread please [click here]

 

Author A better direct sparse solver for linear system with a positive definite matrix
yuwb

2006-05-26, 7:04 pm

I am searching for a better direct sparse solver for linear system with
a positive definite matrix. I am currently using skyline storage along
with cholesky decomposition. Is there a better alogrithm?

Victor Eijkhout

2006-05-28, 8:04 am

yuwb <wenbin.yu@usu.edu> wrote:

> I am searching for a better direct sparse solver for linear system with


Better than what?

> a positive definite matrix. I am currently using skyline storage along
> with cholesky decomposition. Is there a better alogrithm?


Permute your matrix first with multiple minimum degree or so.

Personally, I would use any of a number of free packages (SuperLU,
Mumps, Spooles, Umfpack, .... Saves you mountains of coding, and it's
most likely going to be ten times faster than you could do anyway.

Victor.
--
Victor Eijkhout -- eijkhout at tacc utexas edu
ph: 512 471 5809
yuwb

2006-05-30, 4:04 am

I used RCM to reorder the matrix. I want it better than the skyline
storage + modified cholesky decomposition without sqare rooting the
diagonal terms.

yuwb

2006-05-30, 4:04 am

Tim,

Can you point to some of these documents? Thanks!

Tim Prince wrote:
> yuwb wrote:
> As some of the more evident answers to this question are well
> documented, I must assume you are rejecting them as a matter of opinion.
> Or, does your spelling of that last word have a tricky meaning?


Sponsored Links







Also available: Server administration forum archive | Web Design forum archive | Software forum archive | Hardware reviews archive

Copyright 2008 codecomments.com