[gmx-users] gromacs.org_gmx-users Digest, Vol 128, Issue 10

[wh]

> >> Dear Gromacs Users,
>
> >>
>
> >>
>
> >>
>
> >> Recently when I tried to restart a md simulation from previous one
>
> >> using convert-tpr with version 5.0.1 , the task failed giving the 
> >> error
> message:
>
> >>
>
> >>
>
> >>
>
> >> X particles communicated to PME node Y are more than a cell length 
> >> out
>
> >> of the domain decomposition cell of their charge group
>
> >>
>
> >>
>
> >>
>
> >> I tried several way to solve it, including modify the equilibration
>
> >> setting as it said in the gromacs error page, and all failed. If I
>
> >> just run one simulation for a long time, it will not crash, at 
> >> least
>
> >> can do further than the time point it crashed when using restart.
>
> >>
>
> >> Does anyone have any idea why this happen and how to solve it?  I 
> >> used
>
> >> the
>
> >> syntax:
>
> >>
>
> >> gmx convert-tpr -s input.tpr -f input.trr -e input.edr -o 
> >> output.tpr
>
> >> -extend
>
> >> 10000
>
> >>
>
> >> Any suggestion will be greatly appreciated
>
> >>
>
>
>
> >Do not supply .trr and .edr files; just increase the number of steps 
> >and
> continue from the existing checkpoint with mdrun -cpi.  What you're 
> doing is based on ancient syntax that is no longer relevant.  I'm 
> guessing that you're probably winding up with discontinuous states 
> somehow.
>
>
>
> >
> http://www.gromacs.org/Documentation/How-tos/Extending_Simulations#Ver
> sion_
> 4_and_Newer
>
>
>
> >-Justin
>
>
>
> Thank you so much for your timely reply, yet I tried mdrun -s 
> input.tpr -cpi input.cpt -append
>
> It still gave the same error message. Could it be a bug? Or I did 
> something wrong?


Bug is possible but unlikely. We definitely don't have enough information to
tell because we've gotten half of it each time - what was the conversion
+ restart combination you actually used? Does the original run stay 
+ stable
if it is not interrupted?

Mark






If I run a long time simulation continuously it can stay stable. I just want
to run it in smaller jobs or restart it from a check point because of the
limit on computing resources. Suggested by  Justin I used syntax:
mdrun -s input.tpr -cpi input.cpt -append, using the cpt file from the
previous simulation yet it still gave the same error message.