[gmx-users] Simulation runs indefinitely

[Mark Abraham]

Hi,

You extended the run time of the tpr file, but mdrun only knows it has a
tpr that asks for a number of steps, and that the input checkpoint file
starts at whatever step it does. If you give it a checkpoint that is from
much earlier in the run, then there's no way for it to know you intended
something else.

Mark

On Wed, 27 Jan 2016 14:36 anu chandra <anu80125 at gmail.com> wrote:

> Many thanks for the reply. I just created tpr file once again to see the
> output message. But, it looks like convert-tpr showing something different
> from it suppose to write about the steps. The only difference in the
> previous run was that it carried out with single GPU node.
>
>
> *******************************************************************************************************
> GROMACS:      gmx convert-tpr, VERSION 5.0.5
> Executable:   /usr/local/Cluster-Apps/gromacs/impi/5.0.5/bin/gmx_mpi
> Library dir:  /usr/local/Cluster-Apps/gromacs/impi/5.0.5/share/gromacs/top
> Command line:
>   gmx_mpi convert-tpr -s 18md.tpr -f 18md.cpt -e 18md.edr -o 19md.tpr
> -extend 20000
>
> Reading toplogy and stuff from 18md.tpr
> Reading file 18md.tpr, VERSION 5.0.5 (single precision)
>
> NOTE: Reading the state from trajectory is an obsolete feature of gmx
> convert-tpr.
>       Continuation should be done by loading a checkpoint file with mdrun
> -cpi
>       This guarantees that all state variables are transferred.
>       gmx convert-tpr is now only useful for increasing nsteps,
>       but even that can often be avoided by using mdrun -maxh
>
> NOTE: The simulation uses pressure coupling and/or stochastic dynamics.
> gmx convert-tpr can not provide binary identical continuation.
> If you want that, supply a checkpoint file to mdrun
>
>
> READING STATE FROM CHECKPOINT 18md.cpt...
>
> Extending remaining runtime of by 20000 ps (now 17529120 steps)
> Writing statusfile with starting step  138970880 and length   17529120
> steps...
>                                  time 277941.760 and length  *35058.240 ps*
>
> gcq#333: "History has expired" (PubMed Central)
>
> ***************************************************************************************************
>
> It written *35058.240 ps *instead of *20000ps. *I usually use mdrun as '
> -cpi 18md.cpt -deffnm 19md '
>
>
> Many thanks
> Anu
>
> On Wed, Jan 27, 2016 at 12:47 PM, Mark Abraham <mark.j.abraham at gmail.com>
> wrote:
>
> > As suggestion suggests, you probably didn't do what you think you did,
> > particularly with restarting from the checkpoint. Do a diff on your log
> > files to see what was different.
> >
> > Mark
> >
> > On Wed, Jan 27, 2016 at 1:46 PM Justin Lemkul <jalemkul at vt.edu> wrote:
> >
> > >
> > >
> > > On 1/27/16 6:28 AM, anu chandra wrote:
> > > > Hello Gromcas users,
> > > >
> > > > I having been simulating few membrane-protein systems for about
> ~300ns,
> > > > using a 20 ns window for each run. Though I used the command '
> gmx_mpi
> > > > convert-tpr -s md20.tpr -f md20.cpt -e md20.edr -o md21.tpr -extend
> > > 20000 '
> > > > to generate the tpr file for the next run, interestingly, one of the
> > > system
> > > > ran for ~26000 ps and only got terminates once the run reached
> > wall-clock
> > > > time. All other systems terminates after successfully completing
> 20000
> > > ps,
> > > > as prescribed.  I just wonder why simulation of one of the system
> > > continued
> > > > indefinitely, as rest of the systems behaved as expected. All systems
> > > have
> > > > almost same composition, just differ in the ion concentration.
> > > >
> > >
> > > There must be some problem in your script when you're creating the
> input.
> > > mdrun
> > > doesn't decide to run any number of steps it likes.  Probably you've
> > > failed to
> > > supply a checkpoint file, so the run starts over from the beginning in
> > that
> > > instance.
> > >
> > > -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
> > >
> > > ==================================================
> > > --
> > > Gromacs Users mailing list
> > >
> > > * Please search the archive at
> > > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> > > posting!
> > >
> > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > >
> > > * For (un)subscribe requests visit
> > > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> > > send a mail to gmx-users-request at gromacs.org.
> > >
> > --
> > Gromacs Users mailing list
> >
> > * Please search the archive at
> > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> > posting!
> >
> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >
> > * For (un)subscribe requests visit
> > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> > send a mail to gmx-users-request at gromacs.org.
> >
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>