Re: [AMBER] About SMD simulation restart file

From: Chinh Su Tran To <chinh.sutranto.gmail.com>
Date: Mon, 17 Oct 2011 13:12:01 +0800

Dear Adrian,

I got a similar situation here. I was running unfolding for my protein. It
was supposed to run for 60ns, but it stopped at the 18th nano-second because
the rst file was not generated at that point of time.

As I checked the results, it shows:

heat3.mdcrd *with the size *1319492081 (~1.3 Gb)
heat3.out 1047322
heat3.rst 1982053

Note that the size are the same as the heat2's files accordingly. But the
heat2 was fine (i.e. heat2.mdcrd has 2000 frames, heat2.rst has 1 frame).
Then I visualized the heat3.mdcrd and heat3.rst on VMD. Only the heat3.mdcrd
showed up (with 2000 frames), but the heat3.rst did not (0 frame).
The VMD error said:

Problem when reading the CRD file

We doubted that the problem came from MPI of sander because it showed "the
mpi exit without calling finalize" when the job ran on 4 nodes (8 ppn).
Hence, I re-ran the job only on 1 node. There was no such error this time,
but still nothing was shown in the heat3.rst!!!!

Could you please give me any suggestion? I am using Amber 10.

Thank you.

Regards,
Chinh

On Thu, Oct 13, 2011 at 4:48 AM, Adrian Roitberg <roitberg.qtp.ufl.edu>wrote:

> yes, absolutely
> now, I am not sure what you mean by 'not usable'
>
> if you mean that data is missing or corrupted, then either the file was
> not fully written when you hit the wallclock
> or that your molecular is flying outside your box (if you see *** for
> instance).
>
> Now, can you visualize the restart file into vmd ?
>
> if that does not work, then the file is truncated/corrupted.
>
> If you can visualize it, try running a regular sander/pmemd run without
> smd. If that works, then the restart is fine...
>
> Remember that to restart an SMD run you MUST change the restraint file,
> since you are now starting your SMD from whatever coordinates you last
> saved !
>
> Adrian
>
>
> On 10/12/11 4:39 PM, Mo Chen wrote:
> > Hi Adrian,
> > I understand that. I actually meant to say that I heard the restart file
> of
> > SMD simulation is not quite usable even if my simulation did not run out
> of
> > the wall time. Basically I have just heard that, but I am sorry I have
> not
> > tried myself and I did not know why. In your last email, do you mean if
> the
> > restart file is saved properly before the end of the walltime, I should
> be
> > able to use it as a restart file? Thank you very much!
> >
> > Best,
> > Mo
> >
> > On Wed, Oct 12, 2011 at 4:30 PM, Adrian Roitberg<roitberg.qtp.ufl.edu
> >wrote:
> >
> >> This has nothing to do with SMD runs.
> >>
> >> If you hit the wallclock time, you sometimes do not see the 'last'
> >> restart fully written.
> >>
> >>
> >> You can pick the last frame form the mdcrd and create your new restart.
> >>
> >> Adrian
> >>
> >>
> >> On 10/12/11 4:26 PM, Mo Chen wrote:
> >>> Dear Amber users/developers,
> >>> I am trying to run SMD simulations on Ranger, but my simulations could
> >> not
> >>> finish before the maximum runtime (24 hours) even though I would use
> the
> >>> optimal number of cores. And I heard the SMD restart file is not quite
> >>> usable, but I did not understand why. May I ask anyone has encountered
> >>> similar situations? And how should I solve this problem? Thank you very
> >>> much!
> >>>
> >>>
> >>> Best,
> >>> Mo Chen
> >>> Graduate Student
> >>> 120 Stocking Hall
> >>> Cornell University
> >>> Ithaca, NY 14850
> >>> _______________________________________________
> >>> AMBER mailing list
> >>> AMBER.ambermd.org
> >>> http://lists.ambermd.org/mailman/listinfo/amber
> >>>
> >>
> >> --
> >> Dr. Adrian E. Roitberg
> >> Professor
> >> Quantum Theory Project, Department of Chemistry
> >> University of Florida
> >> roitberg.ufl.edu
> >>
> >> _______________________________________________
> >> 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
> >
>
> --
> Dr. Adrian E. Roitberg
> Professor
> Quantum Theory Project, Department of Chemistry
> University of Florida
> roitberg.ufl.edu
>
> _______________________________________________
> 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 Sun Oct 16 2011 - 22:30:03 PDT
Custom Search