[gmx-users] mdrun rerun to large frame numbers

[Mark Abraham]

Hi,

On Thu, Jul 7, 2016 at 3:17 PM Sebastian <
sebastian.buchenberg at physik.uni-freiburg.de> wrote:

> Hi,
>
> On 07/07/2016 12:53 PM, Mark Abraham wrote:
> > Hi,
> >
> > Hmm, that sounds like your very long trajectory is running into some
> > integer exceeding its maximum size. mdrun proper uses a 64-bit integer
> for
> > step number, but doubtless there are dusty corners. Are you able to
> > estimate which frame number this was?
>
> It seems to be a problem of the step number (around 2e+9) to make the
> trouble.
>

A signed 32-bit integer indeed has a range up to +2e+9, so that confirms my
suspicion.

> What does gmx check say about the
> > very long trajectory?
>
> gmx check works fine. Giving all the outputs I am used to like:
>
> Step  500000
> Coords 500000 ........................
>
> no trouble.
>
> > To work around it, you can probably just use gmx
> > trjconv to break the trajectory into parts, but we'd rather find the
> > problem and fix it :-)
>
> I tried this now it always works for the first half but not for the second.
>

Shucks. However, that makes me expect that you can probably reproduce the
problem with a very short piece of trajectory that includes a step number >
2^31 = 2147483648. If so, please mail me that trajectory + .tpr + failing
.log file and I can probably fix it quickly once I find which bit is
failing.

Is there a way to change the step number? Since the frame number does
> not seem to produce the error.
>

I'm not aware of a mechanism, and a quick glance at the code of trjconv and
trjcat didn't suggest anything. I had an idea that gmx trjcat -settime on
the second half might be useful, but I now doubt it will work.

Mark

>
> > Mark
>
>
> Thanks,
>
> Basti
>
>
> >
> > On Thu, Jul 7, 2016 at 11:59 AM Sebastian <
> > sebastian.buchenberg at physik.uni-freiburg.de> wrote:
> >
> >> Hey,
> >>
> >> I try to perform an rerun of very long trajectories using mdrun.
> >> The problem is, that the rerun crashes after it reaches a certain frame
> >> number with the warning:
> >>
> >> there may be something wrong with the energy file XXX.edr
> >> Found: step=-1983077269, nre=49, nblock=0, time=4.62378e+06.
> >> Trying to skip frame expect a crash though.
> >>
> >> I tried it for three different runs and I get always the same warning.
> >> The step number is something strange.
> >>
> >> Best,
> >>
> >> Basti
> >> --
> >> 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.
>