[Bug #902760] Testsuite fails with ARPACK 3.6 (#1107)
Sylvestre Ledru
sylvestre at debian.org
Sun Aug 12 09:31:54 BST 2018
Looks like https://github.com/opencollab/arpack-ng/issues/142 ?
S
On 27/07/2018 01:34, Andreas Tille wrote:
> Hi,
>
> just forwarding the suspicion of igraph upstream that the problem is
> caused by arpack. I'm tempted to reassign bug #902760 but I'd like
> to hear your opinion first.
>
> Kind regards
>
> Andreas.
>
> ----- Forwarded message from Tamás Nepusz <notifications at github.com> -----
>
> Date: Thu, 26 Jul 2018 11:19:03 -0700
> From: Tamás Nepusz <notifications at github.com>
> To: igraph/igraph <igraph at noreply.github.com>
> Cc: Andreas Tille <tille at debian.org>, Author <author at noreply.github.com>
> Subject: Re: [igraph/igraph] Testsuite fails with ARPACK 3.6 (#1107)
>
> Okay, I am pretty certain that this is an inherent issue with ARPACK 3.6; the same problem can also be triggered in GNU Octave (4.4.0) when linked with ARPACK 3.6 (I tried it on a Mac with Homebrew). The code below produces the PageRank matrix of the graph from the igraph test case:
>
> ```
> octave:1> m = ones(11, 11) * 0.15 / 11;
> octave:2> m(1, 2:11) += 0.85;
> octave:3> m(2:11, 1) = (1 - m(1, 1)) / 10;
> ```
>
> and it blows up exactly the same way:
>
> ```
> octave:4> [v, lambda] = eigs(m, 1)
> error: __eigs__: eigs: error -9999 in dnaupd
> error: called from
> eigs at line 285 column 18
> ```
>
> I'll file a bug report to `arpack-ng`.
>
More information about the debian-science-maintainers
mailing list