Re: Exploiting sparse matrices for speed

[Jordi Gutiérrez Hermoso]

On 5 November 2012 16:59, Joza <address@hidden> wrote:
> In light of another thread I had, I thought it might be good to start a more
> general discussion on the topic. As I stated in the other thread, I had a
> 1000x1000 sparse matrix with 2 on the main-diagonal, and -1 and the upper
> and lower sub diagonal. All other entries were zero. Using a Successive
> Over-Relaxation method to solve this took FOREVER in my opinion....to my
> desired accuracy, hours.
>
> Surely there are ways of exploiting such massive sparsity for faster
> algorithms! I'd be shocked if there weren't

I already answered this:

http://www.gnu.org/software/octave/doc/interpreter/Sparse-Linear-Algebra.html

For this particular case, it will be a Cholesky factorisation.

- Jordi G. H.