[gmx-users] Continuation of the gromacs job using gmx convert-tpr

Mark Abraham mark.j.abraham at gmail.com
Fri Jun 22 15:02:42 CEST 2018 

Hi,

There are some differences in recent GROMACS versions here (because the old
implementations were not robust enough), but the checkpoint restart will
not work with appending unless it finds the output files named in the .cpt
match those on the command line (here, from -deffnm). You're making extra
work for yourself in several ways.

I encourage you to not use -deffnm with a new name that merely signifies
that the extension happened. There's no physical and no real organizational
reason to do this.

If you want numbered output files for each step, then start your
simulations with -noappend and let mdrun number them automatically. But IMO
all that does is make work for you later, concatenating the files again.

If you want appending to work after extending to the number of steps, use
-s new.tpr -deffnm old rather than -deffnm new, because the former doesn't
create name mismatches between those output files that the checkpoint
remembers and those you've instructed mdrun to use now.

And if your reason for using -deffnm is that you want to have multiple
simulation steps in the same directory, bear in mind that using a single
directory to contain a single step is much more robust (you are using the
standard way of grouping related files, called a directory, and using cd is
not any more difficult than -deffnm), and you can just use the default file
naming:

(cd step7; mpirun -np whatever gmx_mpi mdrun -s extended)

Mark

On Fri, Jun 22, 2018 at 11:07 AM Own 12121325 <own12121325 at gmail.com> wrote:

> thanks Mark!
>
> could you please also confirm that my method of the prolongation of the
> simulation would be correct
>
> #entend simulation for 50 ns and save these pieces as the separate files
> with the name step7_2*
> gmx convert-tpr -s step7_1.tpr -o step7_2.tpr -extend 50000
> mdrun -v -deffnm step7_2 -cpi step7_1.cpt
>
> 2018-06-22 10:57 GMT+02:00 Mark Abraham <mark.j.abraham at gmail.com>:
>
> > Hi,
> >
> > The previous checkpoint has the _prev suffix, in case there is a problem
> > that might require you to go further back in time.
> >
> > Mark
> >
> > On Fri, Jun 22, 2018, 10:46 Own 12121325 <own12121325 at gmail.com> wrote:
> >
> > > P.S. what the difference between name.cpt and name_prev.cpt produced by
> > > mdrun? What check-point should correspond to the last snapshot in trr
> > file
> > > ?
> > >
> > > 2018-06-22 10:17 GMT+02:00 Own 12121325 <own12121325 at gmail.com>:
> > >
> > > > In fact there is an alternative trick :-)
> > > > If I rename a tpr file via gmx convert-tpr  and then run mdrun using
> > this
> > > > new tpr as well as previous checkpoint, it will produce all pieces of
> > the
> > > > trajectory in the separate file:
> > > >
> > > > gmx convert-tpr -s step7_1.tpr -o step7_2.tpr -extend 50000
> > > > mpirun -np ${NB_TASKS} mdrun -v -deffnm step7_2 -cpi step7_1.cpt
> > > >
> > > > If I add -noappend flag to the mdrun, its also do the same job but
> also
> > > > will add suffix pat002 to each of the new file (that is not necessary
> > for
> > > > me since I have already renamed tpr).
> > > >
> > > > Gleb
> > > >
> > > >
> > > > 2018-06-21 14:17 GMT+02:00 Justin Lemkul <jalemkul at vt.edu>:
> > > >
> > > >>
> > > >>
> > > >> On 6/21/18 2:35 AM, Own 12121325 wrote:
> > > >>
> > > >>> and without append flag it will produce an output in the separate
> > file,
> > > >>> won't it?
> > > >>>
> > > >>
> > > >> No, because appending is the default behavior. Specifying -append
> just
> > > >> invokes what mdrun does on its own. If you want a separate file, add
> > > >> -noappend to your mdrun command.
> > > >>
> > > >> -Justin
> > > >>
> > > >>
> > > >> gmx convert-tpr -s init.tpr -o next.tpr -extend 500000
> > > >>> gmx mdrun -s next.tpr -cpi the_last_chekpoint.cpt
> > > >>>
> > > >>> 2018-06-21 1:12 GMT+02:00 Justin Lemkul <jalemkul at vt.edu>:
> > > >>>
> > > >>>
> > > >>>> On 6/19/18 4:45 AM, Own 12121325 wrote:
> > > >>>>
> > > >>>> Hello Justin,
> > > >>>>>
> > > >>>>> could you specify please a bit more. Following your method, if
> the
> > > >>>>> simulation has been terminated by crash without producing gro
> file
> > so
> > > >>>>> to
> > > >>>>> re-initiate it I only need one command:
> > > >>>>>
> > > >>>>> gmx mdrun -s initial.tpr -cpi the_last_chekpoint.cpt -append
> > > >>>>>
> > > >>>>> where the last_checkpoint should be something like initial.cpt or
> > > >>>>> initial_prev.cpt
> > > >>>>>
> > > >>>>> Right.
> > > >>>>
> > > >>>> but In case if my simulation has been finished correctly e.g. for
> 50
> > > ns
> > > >>>> and
> > > >>>>
> > > >>>>> I now need to extend it for another 50 ns,  should I do the
> > following
> > > >>>>> trick
> > > >>>>> with 2 GMX programs:
> > > >>>>>
> > > >>>>> gmx convert-tpr -s init.tpr -o next.tpr -extend 500000
> > > >>>>> gmx mdrun -s next.tpr -cpi the_last_chekpoint.cpt -append
> > > >>>>>
> > > >>>>> it will produce the second part of the trajectory
> <https://maps.google.com/?q=roduce+the+second+part+of+the+trajectory+&entry=gmail&source=g>as
> the new file
> > > >>>>> (next.trr) or merge together the first and the second part ?
> > > >>>>>
> > > >>>>> You're specifying -append, so the output will be concatenated by
> > > mdrun.
> > > >>>> That's also been default behavior for as long as I can remember,
> too
> > > :)
> > > >>>>
> > > >>>> -Justin
> > > >>>>
> > > >>>> --
> > > >>>> ==================================================
> > > >>>>
> > > >>>> Justin A. Lemkul, Ph.D.
> > > >>>> Assistant Professor
> > > >>>> Virginia Tech Department of Biochemistry
> > > >>>>
> > > >>>> 303 Engel Hall
> > > >>>> 340 West Campus Dr.
> > > >>>> Blacksburg, VA 24061
> > > >>>>
> > > >>>> jalemkul at vt.edu | (540) 231-3129
> > > >>>> http://www.thelemkullab.com
> > > >>>>
> > > >>>> ==================================================
> > > >>>>
> > > >>>> --
> > > >>>> 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.
> > > >>>>
> > > >>>>
> > > >> --
> > > >> ==================================================
> > > >>
> > > >> Justin A. Lemkul, Ph.D.
> > > >> Assistant Professor
> > > >> Virginia Tech Department of Biochemistry
> > > >>
> > > >> 303 Engel Hall
> > > >> 340 West Campus Dr.
> > > >> Blacksburg, VA 24061
> > > >>
> > > >> jalemkul at vt.edu | (540) 231-3129
> > > >> http://www.thelemkullab.com
> > > >>
> > > >> ==================================================
> > > >>
> > > >> --
> > > >> 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.
> >
> --
> 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.
>

More information about the gromacs.org_gmx-users mailing list