[gmx-users] System blows up

[Mark Abraham]

On 11/01/2012 8:35 PM, Guido Leoni wrote:
> Dear List
> I'm newbie to gromacs (sigh). I'm simulating the interaction between 
> two proteins(chrystallographic structures) following the spider toxin 
> tutorial. I'm able to correctly minimize ( E<1000) and equilibrate 
> (400 ps ) the system. No error message appears to me and I'm quite 
> sure that my starting structures have no steric clashes.
> My .mdp files are equal to the tutorial description excepted for the 
> choosen timesteps (400 ps in equilibration and 1000 ps in dynamics).
> When I run the dynamic (launching gromacs with mpi) I retrieve after 7 
> ps the following error message
> "2 particles communicated to PME node 3 are more than a cell length 
> out of the domain decomposition cell of their charge group"
> I suppose that this is due to the energy of some components that blow up.
> Please how can i check which components are responsible for  the problem?
> I'm also looking the instruction on the web site but they appear to me 
> a bit cryptic. For example
>
> If the crash is happening relatively early (within a few steps), set 
> nstxout (or nstxtcout) to 1, capturing all possible frames.  Watch the 
> resulting trajectory to see which atoms/residues/molecules become 
> unstable first.
>
> I looked the resulting trajectories of the 2 proteins with ngmx but 
> they seem to be always in the box and it is impossible to graphically 
> check the water molecules .
>

The next piece of advice is to simplify your problem. If you can get 
each of your proteins to equilibrate on their own, then you probably 
have an atomic clash after all. If one doesn't equilibrate, then maybe 
your simulation conditions or topology is suspect.

Mark