[gmx-users] PAUSE/CONTINUE A JOB

[Anton Feenstra]

David van der Spoel wrote:

> On Fri, 26 Dec 2003, Raj Badhan wrote:
> 
> 
>>Hi All,
>>Hope you all had a good Xmas.
>>Just a quick question. Is there any way to pause a job after a 
>>particular time and then continue the run later? Or could I run a 
>>simulation for a certain amount of time, then use the output to 
>>carry on the run later on.
>>This is mainly for the purposes of running parallel on a 
>>supercomputer. We only have a 2 day maximum run limit, and 
>>want to know if we could pause/stop and recontinue?
> 
> 
> yes, you can use tpbconv to make a continuation tpr file. From 3.2.0 
> onwards the continuation will be binary identical to the original run, 
> even with temperature coupling

You do need to write coordinates and velocities to a .trr file. I usually
set nstxout = 5000 and nxtvout = 5000. That way, I get a frame in the
.trr file only each 100 ps, but that is OK for restarting. You can also
send a 'USR1' signal to mdrun (e.g. with 'killall -USR1 mdrun', or
'kill -USR1 <pid>', the pid is printed in the first line of the md*.log
files). Then, mdrun will stop right after writing the next frame to .trr.


-- 
Groetjes,

Anton
  _____________ _______________________________________________________
|             |                                                       |
|  _   _  ___,| K. Anton Feenstra                                     |
| / \ / \'| | | Dept. of Pharmacochem. - Vrije Universiteit Amsterdam |
|(   |   )| | | De Boelelaan 1083 - 1081 HV Amsterdam - Netherlands   |
| \_/ \_/ | | | Tel: +31 20 44 47608 - Fax: +31 20 44 47610           |
|             | Feenstra at chem.vu.nl - www.chem.vu.nl/~feenstra/       |
|             | "If You See Me Getting High, Knock Me Down"           |
|             | (Red Hot Chili Peppers)                               |
|_____________|_______________________________________________________|