Re: [Help-gsl] svd

[Maxime Boissonneault]

Hi Philippe,
If the matrix is dense, LAPACK is your best bet. I am not sure it allows 
to compute only a few singular values though. You can also check ARPACK, 
which definitely allows you to compute only one or a few singular 
values, but works on sparse matrices (maybe also on dense, but I've only 
ever used it for sparse ones).

Best,

Maxime Boissonneault

Le 12-02-03 04:02, philippe preux a écrit :
> Hi,
> I would like to compute only a given number of singular values (n 
> largests), and their associated singular vectors.
> Is there any means to do that with GSL?
> Is there any means to compute only the values, not the vectors?
> If not, is there any way to do it with standard packages (lapack, 
> ...)? (sorry, this is not a GSL only question, but I guess it is 
> relevant though).
> My goal is to handle (rather) as large matrices as possible (I hope 
> many thousands rows and lines) which happen to be square (and 
> symmetric), within reasonnable computational times on my laptop.
> Thanks for any help,
> Philippe
>
> _______________________________________________
> Help-gsl mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/help-gsl