[gmx-users] gmx covar (Version 5.x) segfaults with covariance matrices bigger than 1500x1500

[Machtens, Jan-Philipp]

Hi,
yes I will post that to redmine soon, but let me first clarify the issue, since I tested some more conditions in the meantime.
(I realized that my trajectory had 1501 frames and the error occured when I choose more than 500 atoms, i.e. when the number of degrees of freedom becomes larger then the number of frames).
So in general, if your trajectory has nf frames, g_covar will result in a segfault if the group to be (covariance) analyzed is larger then nf/3 atoms. 
If found this statement to be true for gromacs 4.6.6, 4.6.7 and 5.0.2.
This error does not happen for gromacs 4.6.5

So, I wonder whether this is an unwanted behaviour which has been introduced in 4.6.6? Do you think this behavior makes sense ? The official release notes for 4.6.6 do not mention changes in g_covar.
Best,
Jan-Philipp Machtens
________________________________________
From: gromacs.org_gmx-users-bounces at maillist.sys.kth.se [gromacs.org_gmx-users-bounces at maillist.sys.kth.se] on behalf of Mark Abraham [mark.j.abraham at gmail.com]
Sent: Thursday, November 06, 2014 1:27 PM
To: Discussion list for GROMACS users
Subject: Re: [gmx-users] gmx covar (Version 5.x) segfaults with covariance matrices bigger than 1500x1500

Hi,

IIRC some minor bugs were fixed, but a regression like that should be dealt
with... please file at http://redmine.gromacs.org.

Mark

On Thu, Nov 6, 2014 at 12:00 PM, Machtens, Jan-Philipp <
j.machtens at fz-juelich.de> wrote:

> Dear all,
> I observed that g_covar in Gromacs 5.0.2 segfaults at the diagonalization
> step, if the group for the covariance analysis is bigger than 500 atoms,
> i.e. if the covariance matrix is bigger than 1500x1500.
> In my hands, this problem does not arise with g_covar 4.6.x. The results
> obtained from g_covar (version 5.x) on groups <= 500 atoms that I obtained
> were the same os those from version 4.6.x and also verified manually.
> Memory is not the problem here. The problem is independent on single/double
> precision.
>
> Before posting this on redmine.gromacs I first want to check whether
> someone here has made a contradicting experience which would argue against
> a possible bug introduced in 5.0.2?
> Thanks.
> Cheers,
> Jan-Philipp Machtens
>
>
>
> ------------------------------------------------------------------------------------------------
>
> ------------------------------------------------------------------------------------------------
> Forschungszentrum Juelich GmbH
> 52425 Juelich
> Sitz der Gesellschaft: Juelich
> Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498
> Vorsitzender des Aufsichtsrats: MinDir Dr. Karl Eugen Huthmacher
> Geschaeftsfuehrung: Prof. Dr.-Ing. Wolfgang Marquardt (Vorsitzender),
> Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt,
> Prof. Dr. Sebastian M. Schmidt
>
> ------------------------------------------------------------------------------------------------
>
> ------------------------------------------------------------------------------------------------
>
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>
--
Gromacs Users mailing list

* Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!

* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists

* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-request at gromacs.org.