[gmx-users] Peptide - DMPC membrane simulations -> Unstable system

[Kirill Bessonov]

Hello gain Justin,

Now I figured out that the problem was in gromacs version 4.0. So peptide in
water works no problem.

I have used trjconv -pbc mol -ur compact and it solved the artifact problem.
Thanks a lot!

Now I have started simulation on 96 CPU's for 1000ns total run time. But
after 1 day simulation the system crashed or exploded with the following
error: [First time it failed because of some problems with the server, and
then I have resumed the run using tpbconv -s DMPCRun1000ns.tpr -f
1000nsDMPCA141Run_part1.trr -extend 100000 -o DMPCRun100nspart2.tpr]

vol 0.78  imb F  2% pme/F 2.15 step 10578400, will finish Sat Aug 29
07:23:01 2009
vol 0.77  imb F  3% pme/F 2.15 step 10578500, will finish Sat Aug 29
07:28:32 2009

Step 10578599, time 21157.2 (ps)  LINCS WARNING
relative constraint deviation after LINCS:
rms 0.299801, max 6.073617 (between atoms 9449 and 9450)
bonds that rotated more than 30 degrees:
 atom 1 atom 2  angle  previous, current, constraint length
   9451   9468   93.6    0.1530   0.1293      0.1530

Fatal error:
15 particles communicated to PME node 6 are more than a cell length out of
the domain decomposition cell of their charge group

I am attaching my mdp files used to to EM and MDrun here:

EM: http://kbessonov.googlepages.com/em_lipid.mdp MD:
http://kbessonov.googlepages.com/mdonly_1000ns.mdp

I was wondering why my system fails. and what can I do to avoid it? I was
thinking maybe to run several times EM on the system, maybe it will help?
What bugs me is that it fails really late in simulation. I have not seen
trajectory file yet.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20090718/888eb9ea/attachment.html>