[gmx-users] g_msd segmentation fault

[Dommert Florian]

On Mon, 2012-03-05 at 23:56 +1100, Mark Abraham wrote: 
> On 5/03/2012 11:39 PM, Gavin Melaugh wrote:
> > Dear All
> >
> > I have a system of 40 solute molecules in 480 crown ether solvent molecules.
> > When I ran the msd analysis on the solvent molecules using the following
> > comand.
> >
> > g_msd_login_d -f traj.trr -s topol.tpr -mol -o 1_12_400K_sol_msd.xvg
> >
> > I get  a segmentation fault as follows
> >
> > Select a group: 4
> > Selected 4: 'SOL'
> > Split group of 16800 atoms into 480 molecules
> > trn version: GMX_trn_file (double precision)
> > Reading frame     800 time 40000.000   Segmentation fault
> >
> >
> > However when I perform the same analysis on the solute molecules it runs
> > to completion. I  have checked the configuration of the system at frame
> > 800 and everything seems to be fine. I have also analysed the energy and
> > there seems to be no problem. Has anybody any idea of what might be
> > happening?
> 
> Running out of memory would be the first suggestion. A segmentation 
> fault is a generic error message when a C program (ie. GROMACS) has 
> assumed that it can get access to a certain amount of memory, but is in 
> fact doing so in a way that breaks the rules that "segment" memory so 
> that one (part of a) program can't hurt another one.
> 
> Running in single precision will roughly halve your memory requirements, 
> probably without affecting your results. Reducing the number of frames 
> can help also, but it seems you already have fairly few frames. Using 
> -mol probably increases the amount of memory you need by an amount 
> proportional to the number of such molecules (or maybe its square?).
> 
> Mark

However, it seems that the mol flag of g_msd is still buggy as described
by: 

http://redmine.gromacs.org/issues/774

Unfortunately, this bug is still not fixed, though a patch seems to be
provided. My workaround for this problem is to convert the trajectory
with trjconv using the flag -pbc whole and apply g_msd without the -mol
option, but providing appropriate index groups for every single
molecule. Unfortunately, this method is very time consuming, because the
trajectory has to be read, many many times. Afterwards, I just average
the results for the different molecules and derive the diffusion
coefficient from the averaged MSD.

/Flo 


-- 
Florian Dommert
Dipl. - Phys.

Institute for Computational Physics
University Stuttgart

Pfaffenwaldring 27
70569 Stuttgart

EMail: dommert at icp.uni-stuttgart.de
Homepage: http://www.icp.uni-stuttgart.de/~icp/Florian_Dommert

Tel.: +49 - (0)711 - 68563613
Fax.: +49 - (0)711 - 68563658
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20120305/9b5a7a48/attachment.sig>