Prev: save multiple simulations data in workspace one simout
Next: peak detection in (x,y,z) data
From: Stonerain on 25 Sep 2009 13:18
Hi everyone,

I have a large (like 10000x10000) sparse matrice and I need to find
that matrice's eigenvalues and eigenvectors. I read eigs is suitable
for this situation but I'm always taking that error "dnaupd did not
find any eigenvalues to sufficient accuracy". How can I get over this
problem?
Any response will be appreciated

Best Regards
From: Nicolas on 2 Oct 2009 16:44
Hi Stonerain,

Maybe the solution is to "decouple" your matrix.
If you break your matrix in small matices, then the spectrum of your biggest matrix is the union of the spectra, isn't it?

Moreover, I think balance() could deal with the eigenvalues accuracy.

I am curious, how precisely do you face this problem?

Regards,
Nico
From: Bruno Luong on 2 Oct 2009 17:10
"Nicolas " <cusseani(a)ensieta.fr> wrote in message <ha5omi$hpe$1(a)fred.mathworks.com>...
> Hi Stonerain,
>
> Maybe the solution is to "decouple" your matrix.
> If you break your matrix in small matices, then the spectrum of your biggest matrix is the union of the spectra, isn't it?
>

Generally false if the smaller ones are not block diagonals of the big one.

Bruno
 | 
Pages: 1
Prev: save multiple simulations data in workspace one simout
Next: peak detection in (x,y,z) data