Re: [AMBER] checkpoint on Amber within batch job

From: Zhuang, Yu <yu.zhuang.tum.de>
Date: Thu, 23 Jan 2014 13:57:44 +0000

Hi,
Thanks for your information. Does this means I just need to re-format the sander command for the next run without modification of .in file? For example,
Original:
srun_ps sander.MPI -O -i 2q2m_wat_md2.in -o 2q2m_wat_md2.out -c 2q2m_wat_md1.restrt -p 2q2m_wat.prmtop -r 2q2m_wat_md2.restrt -x 2q2m_wat_md2.mdcrd
step 2 simulation:
srun_ps sander.MPI -O -i 2q2m_wat_md2.in -o 2q2m_wat_md2.out -c 2q2m_wat_md2.restrt -p 2q2m_wat.prmtop -r 2q2m_wat_md3.restrt -x 2q2m_wat_md3.mdcrd
once again, thanks. Yu
-----Ursprüngliche Nachricht-----
Von: Jason Swails [mailto:jason.swails.gmail.com]
Gesendet: Thursday, January 23, 2014 2:44 PM
An: amber.ambermd.org
Betreff: Re: [AMBER] checkpoint on Amber within batch job

On Thu, 2014-01-23 at 13:33 +0000, Zhuang, Yu wrote:
> Dear all,
> I am wondering how to make a checkpoint on Amber 12 within batch job. I am running Amber on Linux cluster, however, they have a time limit for 48 hours. I would like to continue or restart the simulation after it stopped. The cluster is running under SLURM system. if anyone has a similar experience, it will be highly appreciated if you can share with me.
> Thanks and regards, Yu
>
> The following is the command I used for sander.MPI.
>
> #!/bin/bash
> #SBATCH -o /home/.../myjob.%j.%N.out
> #SBATCH -D /home/.../
> #SBATCH -J beta_lg_1
> #SBATCH --get-user-env
> #SBATCH --clusters=MPP1
> #SBATCH --ntasks=32
> #SBATCH --partition=myri_large
> #SBATCH --time=48:00:00
>
> source /etc/profile.d/modules.sh
>
> module load amber
>
> srun_ps sander.MPI -O -i 2q2m_wat_min1.in -o 2q2m_wat_min1.out -c
> 2q2m_wat.inpcrd -p 2q2m_wat.prmtop -r 2q2m_wat_min1.restrt -ref
> 2q2m_wat.inpcrd; srun_ps sander.MPI -O -i 2q2m_wat_min2.in -o
> 2q2m_wat_min2.out -c 2q2m_wat_min1.restrt -p 2q2m_wat.prmtop -r
> 2q2m_wat_min2.restrt; srun_ps sander.MPI -O -i 2q2m_wat_md1.in -o
> 2q2m_wat_md1.out -c 2q2m_wat_min2.restrt -p 2q2m_wat.prmtop -r
> 2q2m_wat_md1.restrt -x 2q2m_wat_md1.mdcrd -ref 2q2m_wat_min2.restrt;
> srun_ps sander.MPI -O -i 2q2m_wat_md2.in -o 2q2m_wat_md2.out -c
> 2q2m_wat_md1.restrt -p 2q2m_wat.prmtop -r 2q2m_wat_md2.restrt -x
> 2q2m_wat_md2.mdcrd

There is no true 'checkpoint' facility in Amber. The solution is to use the restart file from the previous simulation to start the next simulation. This is almost always as good as a checkpoint file.

HTH,
Jason

--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jan 23 2014 - 06:00:04 PST
Custom Search