Re: [ublas] eigenvalue solvers ... again

From: Nico Galoppo (ngaloppo_at_[hidden])
Date: 2007-11-13 20:02:38

Riccardo,

On Oct 22, 2007 3:12 AM, Riccardo Rossi <rrossi_at_[hidden]> wrote:
Riccardo,

surprisingly, I have a need for an eigenvalue solver myself all of a
sudden. Have you succeeded in making some kind of native ublas
implementation?

Thanks,

--nico

> Hi All,
>
> thanks for your attention ... we will be trying to develop our own
> version in pure ublas ... we will let you know if we have any success
>
> greetings
> Riccardo
>
>
> Nico Galoppo wrote:
> > I've been told that ARPACK++ takes care of wrapping the reverse
> > communication for you.
> >
> > On 10/19/07, Karl Meerbergen <Karl.Meerbergen_at_[hidden]> wrote:
> >
> >> I don't. Actually, a bindings to BLZPACK would come in handy. Note that
> >> arpack is not designed for buckling and frequency analysis unless the
> >> number of wanted eigenvalues is small. It does not do the Sturm sequence
> >> test either. Blzpack also uses reverse communication as arpack. I have
> >> some ideas to make nice bindings to arpack, but no time at the moment ...
> >>
> >> Karl
> >>
> >>
> >> Riccardo Rossi wrote:
> >>
> >>
> >>> i was interested in sparse... :-(
> >>>
> >>> do you have by chance any example using either UMFPACK or Arpack?
> >>>
> >>> thanks in advance
> >>> Riccardo
> >>>
> >>>
> >>>
> >>> Nuno Sucena Almeida wrote:
> >>>
> >>>
> >>>
> >>>>      Dense or sparse matrices? For dense I think you can use Lapack with some
> >>>> bindings laying around somewhere, if you can't find them I think I have them.
> >>>>      For sparse and with Arpack it's more trickier because of the callback
> >>>> mechanism, so what I use is umfpack along the way too.
> >>>>
> >>>>                                                              Nuno
> >>>>
> >>>> On Tuesday 16 October 2007, Riccardo Rossi wrote:
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>> "native ublas" implementation of Lanczos or of some other eigenvalue
> >>>>> solver?
> >>>>> I am interested in generalized eigenvalue solvers to be used in buckling
> >>>>> and frequency analysis
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>
> >>>>
> >>> _______________________________________________
> >>> ublas mailing list
> >>> ublas_at_[hidden]
> >>> http://lists.boost.org/mailman/listinfo.cgi/ublas
> >>>
> >>>
> >>>
> >> _______________________________________________
> >> ublas mailing list
> >> ublas_at_[hidden]
> >> http://lists.boost.org/mailman/listinfo.cgi/ublas
> >>
> >>
> >>
> >>
> >
> >
> >
>
> _______________________________________________
> ublas mailing list
> ublas_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/ublas
>
>

-- 
Nico Galoppo :: http://www.ngaloppo.org