[gmx-users] segmentation fault

[Marc F. Lensink]

On Tue, Dec 11, 2007 at 12:44:28PM +0100, Magnus Andersson wrote:
> Hi all,
> 
> My constrained 20ps simulation crashes for some reason. This simulation is
> very much alike other simulations I've run in a successful manner
> previously. Seems like there is a probem on atoms constituting an
> artificial membrane...
> 
> Anyway, I guess it is a general question. Where do you start looking if
> the run crashes like this (and a slightly different system, just strctural
> change, it has exactly the same atoms...)?

sometimes the system is caught in a deep and steep energy well with
wide barriers so that the assignment of new velocities also doesn't
work.  if the system is constructed from a previous simulation, find a
few time frames before and re-construct.  otherwise i suppose you
could identify the faulty atom and delete the nearest water molecule.

cheers,
marc


-- 
Marc F. Lensink
Centre for Structural Biology and Bioinformatics         SCMBB
Université Libre de Bruxelles          lensink at scmbb.ulb.ac.be
Boulevard du Triomphe - CP 263, B-1050 Brussels, Belgium
tel: +32 2 650 5411  secr: +32 2 650 2013  fax: +32 2 650 5425