[gmx-users] Dividing an MD run

[Justin Lemkul]

On 3/3/16 5:23 PM, Guillem Prats Ejarque wrote:
> Hi,
>
>
> I have to do a production MD in an HPC cluster and due to the company's policy I have to resubmit the job in the queue everyday. This is probably a stupid question, but, to resubmit it starting from the point where I left it, do I only have to run the MD again with the output files of the previous one, like this?
>
>
> grompp -f remd.mdp -c md.gro -t md.cpt -p topol.top -n index.ndx -o remd.tpr
>
> mdrun ....
>
>
> or should I take into account anything else? Will I obtain the same results if I do the dynamic in a single run, or in several steps?
>
>

Once you've created the .tpr file, all you need to do is add -cpi -append to 
your mdrun command to load from the last checkpoint.  Nothing more to it.  You 
can also use -maxh to ensure a clean exit when you hit the wallclock limit.  The 
continuation from the checkpoint is exact.

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================