On Tue, Apr 5, 2011 at 4:26 PM, Martin Helm
<address@hidden> wrote:
Am Dienstag, 5. April 2011, 11:06:09 schrieb nuncio m:
> Hi list,
> What is the maximum size of a matrix that octave can hold. Why I am
> asking this is, I am getting a memory when I try to find the eigen values
> of a matrix of 348 rows and 29898 columns. I am using a 32 bit open suse
> on pentium 4 processor having 2 GB of RAM and a swap of 4.6 GB. Any help
> would be greatly appreciated.
>
> nuncio
Your matrix is fairly small around 80 MB memory consumption, so no problem in
itself at all. The amount of memory you can use depends on your system
architecture. With 32 bit you can use up to 2GB for a matrix, with a 64 bit
(linux) system you can have a matrix up to 16 GB, this limit does not exist if
octave is built with 64 bit indexing enabled (but this is still experimental I
think).
You say you run out of memory calculating the eigen values of a nonsquare
matrix (?). can you show us a short code snippet how you do that? I am not
sure if I really understand.
Do you calculate something like eig(a'*a) or eig(a*a')? If yes you are better
served with looking at the svd command instead of calculating a matrix
product.