[SciPy-Dev] Sparse generalized eigenvalue problems

[Aric Hagberg]

On Fri, Apr 8, 2011 at 6:06 AM, Nils Wagner
<nwagner at iam.uni-stuttgart.de> wrote:
> On Fri, 8 Apr 2011 11:34:58 +0000 (UTC)
>  Pauli Virtanen <pav at iki.fi> wrote:
>>Fri, 08 Apr 2011 10:46:54 +0200, Nils Wagner wrote:
>>> what is recommended to tackle sparse generalized
>>>eigenvalue problems in
>>> scipy?
>>
>> Writing a patch that adds support for generalized
>>eigenvalue problems to
>> eigs :)
>>
>> The necessary low-level ARPACK routines can be found in
>> scipy.sparse.linalg.eigen.arpack._arpack -- in principle
>>you can use
>> those directly, although they shouldn't be treated as
>>public.
>>
>> --
>> Pauli Virtanen
>>
>
> BTW, which version of Arpack is used by scipy ?
>
> I found an interesting comment by Andreas Klöckner. See
>
> http://mathema.tician.de/node/373
>
> for details.
>
> http://mathema.tician.de/software/arpack
>

From
https://github.com/scipy/scipy/blob/master/scipy/sparse/linalg/eigen/arpack/README
---
This is the ARPACK package from
http://www.caam.rice.edu/software/ARPACK/

Specifically the files are from
http://www.caam.rice.edu/software/ARPACK/SRC/arpack96.tar.gz
with the patch
http://www.caam.rice.edu/software/ARPACK/SRC/patch.tar.gz
---

So it doesn't include those patches regarding the "second()"
function/subroutine signatures being different in ARPACK v LAPACK.
But I think we avoid that problem since we explicitly include and
build the ARPACK second() subroutine.

Aric