[gmx-users] on crashed MD

[Justin Lemkul]

On 4/2/16 11:36 PM, Brett wrote:
> Dear All,
>
> Before when I got a MD crashed, I do the following to continue it,
>
> "gmx  tpbconv -s md_0_1.tpr -extend 0 -o md_0_1extending0ps.tpr
> gmx mdrun -deffnm md_0_1extending0ps -cpi md_0_1_prev.cpt -reprod -v &"
>
> Recently I have a crashed MD, and when I do "gmx  tpbconv -s md_0_1.tpr -extend 0 -o md_0_1extending0ps.tpr", it says tpbconv has been replaced by convert-tpr.
>
> In "http://jenkins.gromacs.org/job/Documentation_Nightly_master/javadoc/onlinehelp/gmx-convert-tpr.html#gmx-convert-tpr", there was introduction on  convert-tpr, which says,
>
> "gmxconvert-tpr can edit run input files in four ways.
>
> 1. by modifying the number of steps in a run input file with options -extend, -until or -nsteps (nsteps=-1 means unlimited number of steps)
>
> 2. (OBSOLETE) by creating a run input file for a continuation run when your simulation has crashed due to e.g. a full disk, or by making a continuation run input file. This option is obsolete, since mdrun now writes and reads checkpoint files. Note that a frame with coordinates and velocities is needed. When pressure and/or Nose-Hoover temperature coupling is used an energy file can be supplied to get an exact continuation of the original run.
>
> 3. by creating a .tpx file for a subset of your original tpx file, which is useful when you want to remove the solvent from your .tpx file, or when you want to make e.g. a pure Calpha .tpx file. Note that you may need to use -nsteps-1 (or similar) to get this to work. WARNING: this .tpx file is not fully functional.
>
> 4. by setting the charges of a specified group to zero. This is useful when doing free energy estimates using the LIE (Linear Interaction Energy) method".
>
>
>
>
> My first question is, when I meet a crashed MD, can I continue it by
>
>
> "gmx  convert-tpr -s md_0_1.tpr -extend 0 -o md_0_1extending0ps.tpr

If you're extending by zero, you're not actually extending, so you do not need 
to do this.

> gmx mdrun -deffnm md_0_1extending0ps -cpi md_0_1_prev.cpt -reprod -v &"?
>

You should investigate the source of the crash before trying to continue.  If 
the "crash" was due to a hardware problem, full disk, etc. that's one thing, but 
if it crashed due to instability, it is pointless to try to just brute force 
your way through.

>
>
>
> My second question is, as for the introduction on convert-tpr says to continue a crashed MD by "gmx  convert-tpr" was obsolete, will you please tell me the command by mdrun to directly continue a crashed MD?
>

Because many (many, many...) years ago the only way to continue was a 
combination of grompp or tpbconv with .trr and .edr files.  This information is 
labeled OBSOLETE for a reason.

-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

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