[gmx-users] g_covar & g_anaeig problems

[Sebastian Breuers]

Hey everyone,

searching the list for the answer to a g_covar problem I've found this 
post of Arne who describes the very same problem that I encounter.

I'm observing this problem on a linux cluster with 32GB memory on each 
node. The g_covar_d application was compiled in double precision with an 
intel compiler and works fine up to a number of atoms of around 15408. 
In tests I could find out that above this number +/- 48 atoms I receive 
the following error output:


Calculating the average structure ...
Last frame         50 time 5000.000

Constructing covariance matrix (46368x46368) ...
Reading frame       0 time    0.000   Segmentation fault

Meaning a matrix over around 46224x46224 elements starts to generate 
this problem.

Did anyone encounter the same problem or does anyone has a hint to solve 
this issue?

Kind regards

Sebastian



Am 25.03.2010 10:05, schrieb Arne.Wagner at student.uib.no:
> Hi everyone,
> I've got the following two problems - hopefully someone can help me
> fixing them.
> 1.:
> I have a larger system ~20000 atoms where I performed an MD run and
> now I want to analyze the entropy. I tried a combination of g_covar
> and g_anaeig to get the values.
> Input:
> g_covar -f XXX_md_eq.xtc -n XXX.ndx -s XXX_md_eq.tpr -v XXX_eivec.trr
> -mwa -l XXX_covar.log
>
> If I choose a larger sub-part of the system (approx. > 4000 elements)
> I get the following error message after the average structure is
> calculated:
>
> Constructing covariance matrix ...
> Reading frame       0 time    0.000   Segmentation fault
>
> Is this a result of insufficient memory or is there an error in the
> whole procedure?
>
> 2.:
> In smaller sub-parts I can get the eigenvectors and when I put them
> into g_anaeig,
>
> g_anaeig -v XXX.trr -entropy
>
> I get values for the quasi-harmonical-approximation but
> an 'inf' oder 'nan' for the Schlitter entropy calculation - even with
> additional/maximum input. Why?
>
> Thanks a lot!
> Awa
>


-- 
_____________________________________________________________________________

Sebastian Breuers               Tel: +49-221-470-4108
EMail: breuerss at uni-koeln.de

Universität zu Köln             University of Cologne
Department für Chemie           Department of Chemistry
Organische Chemie               Organic Chemistry

Greinstraße 4                   Greinstraße 4
D-50939 Köln                    D-50939 Cologne, Federal Rep. of Germany
_____________________________________________________________________________