[gmx-users] Fwd: Issue with parallel simulations in vacuum

[Mark Abraham]

Hi,

In GROMACS 4.6.6 we added a fatal error for Andersen+constraints+domain
decomposition with more than one domain, because it gave a temperature that
is obviously too high. So, I presume Michael Shirts never implemented it to
work / tested it with particle decomposition, either.

Fortunately, 600 particles isn't likely to show an appreciable speedup for
more than one rank, so I'd stay on a single rank, or update to a recent
GROMACS and use Verlet + md + v-rescale.

Mark

On Tue, May 16, 2017 at 4:05 PM Justin Lemkul <jalemkul at vt.edu> wrote:

>
>
> On 5/16/17 7:26 AM, Agusti Emperador wrote:
> > Dear GROMACS experts,
> >
> > I am simulating a system of two disordered proteins in vacuum. I use
> > gromacs 4.6.5. The serial simulation works fine, but when I run it in
> > parallel (2 processors) it gives seriously pathological results.
> >
> > I use particle decomposition and periodic boundary conditions, because I
> > want to simulate a solution at a certain concentration (technically, the
> > simulation is in vacuum, but I have included the effects of solvation in
> > the nonbonded terms of the force field in my coarse-grained protein
> model).
> > Each protein has 300 particles.
> >
> > I use exactly the same parameters and input files in the serial
> simulation
> > and in the parallel simulation. This is the mdp file that I use:
> >
> > integrator       =  md-vv
> > dt               =  0.02
> > nsteps           =  500000000
> > nstxout          =  0
> > nstvout          =  0
> > nstenergy = 500000
> > nstlog           =  500000
> > nstxtcout        =  50000
> > nstcomm = 1
> > nstcalcenergy = 1
> > xtc-precision    =  10
> > rlist            =  1.4
> > coulombtype      =  shift
> > rcoulomb         =  1.2
> > epsilon_r        =  15
> > vdw-type         =  shift
> > rvdw-switch      =  0.9
> > rvdw             =  1.2
> > tcoupl           =  andersen
> > tc-grps          =  Protein
> > tau-t            =  .1
> > ref-t            =  300
> > Pcoupl           =  no
> > refcoord_scaling =  all
> >
> > It works perfectly in the serial simulation, but I need to make parallel
> > simulations in order to simulate systems with many proteins.
> >
> > How could I solve the problem?
> >
>
> What is the problem?  What do you define as "seriously pathological
> results" in
> this case?  Also note that 4.6.5 is practically prehistoric and you may
> wish to
> consider a newer, faster, less-buggy version of the code.
>
> -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.
>