[gmx-users] LINCS and parallel simulation

[Philip Shushkov]

Dear gmx-users,

I have a problem concerning LINCS crashes after transfering a simulation
into parallel mode. Due to inherent errors in GROMAS 3.3.2 positional
restraints applied on multiple molecules of the same kind could not be
handled in parallel and I did the initial part of the simulation in serial
mode as suggested by some of the gromacs user. However, due to time
restraints and computational resourses I would like to conduct the rest of
the simulation in parallel without the positional restraints, of course. I
prepared a *.trp file using the -shuffle and -sort options but at the very
beginning of the run the LINCS crashed. My system consists of 81 DPPC
molecules and 8280 water molecules. The errors in the constraints always
appeared to originate from water molecules. I tried many ways to get around
the problem. First, I united the lipids into one molecule but thi did not
help. I reduced the time step to 1 fs or made use of other (not the latest)
points to continue the unrestrained trajectory but all the efforts were in
vain. Also, I switched off the -shuffle and -sort options as I read that
sometimes they could cause run crashes.

Have somebody else encountered such an issue and if so, could he/she propose
how to tackle the problem.

Best wishes,
Philip
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20080505/dfae26e9/attachment.html>