[gmx-users] Can't do parallell free energy calculation with couple_intramol=no -- workaround?

[Szilárd Páll]

On Thu, Mar 10, 2016 at 8:14 PM, Michael Daily <mdaily.work at gmail.com>
wrote:

> Hi,
>
> I am trying to do a solvation free energy calculation on a small peptide in
> gromacs 5.0.  The peptide is in an extended conformation about 2.5 nm long,
> and I use position restraints on the heavy atoms to keep it in this
> conformation.  Due to the nature of the calculation I set
> "couple_intramol=no" in the mdp file (attached for lambda=0).
>
> However, due to the way couple_intramol is implemented in gromacs
> (long-range exclusions), the longest bonded interaction is as long as the
> molecule, leading to a minimum cell size of over 3 nm, which makes domain
> decomposition all but impossible (see the attached log file).  Is there a
> way to limit long-range exclusions and still get reasonable charging free
> energies? I have a few ideas but would like some opinions on whether they
> are physically sound:
>
> 1) change "couple_intramol" to yes and generate intramolecular exclusions
> manually with some cutoff.  Since rlist/rcoulomb/rvdw are 1.2 nm for this
> simulation, I might use a cutoff of 1.5 to provide a buffer.
> 2) Use "energygrp-excl = Protein Protein" to turn off nonbonded
> interactions within the peptide.  The only problem with this is it forces
> the use of group instead of Verlet cutoff scheme which seems to slow
> performance.
> 3) Revert to Gromacs 4.7 and use particle decomposition
>

I guess you mean v4.6.7. As far as I know there is the decomposition should
not interfere with the physics. However, you could instead consider using
5.1 with multi-threading. Unless something is unsupported or you happen to
stumble into code not OpenMP-parallelized, you should be able to run on up
to 24-32 cores without domain-decomposition (also note that 5.1 support
running separate PME without DD).


> The log file from my (attempted) simulation with these parameters is also
> attached.
>

The user's list does not accept attachments, so you'll have to share links
to files uploaded somewhere else.

--
Szilárd



>
> Thanks,
>
> ====================================
> Michael D. Daily
> Research Scientist, Biochemistry and Molecular Biology
> University of Texas Medical Branch at Galveston
> 509-713-0726
>
> --
> 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.
>