[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: eigs failure
From: |
Rik |
Subject: |
Re: eigs failure |
Date: |
Sun, 01 Jul 2012 08:42:31 -0700 |
On 06/30/2012 09:45 AM, address@hidden wrote:
> From: Ben Abbott <address@hidden>
> To: "address@hidden mailing list" <address@hidden>
> Subject: new test failure for eigs.cc in stable branch
> Message-ID: <address@hidden>
> Content-Type: text/plain; CHARSET=US-ASCII
>
> When 3.6.2 was released, I didn't encounter any unexpected test errors. Now
> I'm seeing the error below. I don't see anything in the changelog that looks
> suspicious. Before I start bisecting, does anyone have any suspects?
Ben,
There is a bug in ARPACK, the underlying library for eigs, that causes
segmentation faults occasionally. For a while, we brought the ARPACK code
into the Octave source tree, patched it, and distributed it with Octave.
However, we moved back from this and produced a patched version which is
available here (http://forge.scilab.org/index.php/p/arpack-ng). We moved
the library to be external again so that everyone in the numerical
computing universe could take advantage of the fixes. I'd check and see
what ARPACK you are actually linking against in your build.
--Rik