[gmx-users] box size and neighbor list cut-off

[Mark Abraham]

Hi,

Under the assumptions underlying the drift tolerance, yes. (It's actually
slightly better than advertised, because of the use of discretized
clusters.) At some point, the numerical perturbation from the constraint
implementation dominates, as you can read about in the PDF reference manual
and the paper cited at the above link. IMO at some previous point the error
inherent in the choice of a statically parameterized model dominates, but
YMMV of course. And any attempt to reduce drift (e.g. use of double
precision and/or large buffers) comes at a computational cost that should
be weighed against the smaller number of samples you obtain from your
finite computing resources.

Mark


On Mon, Aug 17, 2015 at 11:08 PM Johnny Lu <johnny.lu128 at gmail.com> wrote:

> Sorry, I mean:
> If I use double precision gmx mdrun and constraint h-bonds, can I expect
> 1.135 rlist to give a drift in total energy as small as the one in the
> table?
>
> On Mon, Aug 17, 2015 at 5:07 PM, Johnny Lu <johnny.lu128 at gmail.com> wrote:
>
> > Thank you for the advices.
> >
> > I tried setting different varlet buffer drift in the 1st equilibration
> run
> > (nvt) for 5 seconds.
> > From the log file of mdrun, I get:
> >
> > rlist        verlet-buffer-drift
> > 1.09        0.000005
> > 1.114        0.0000005
> > 1.135        0.00000005 <=
> >
> > 1.173        0.0000000005
> > 1.234        0.00000000000005
> > 1.284        0.000000000000000005
> >
> >
> > Are these numbers real?
> >
> > The wiki says the constraints already causes 0.0001kj/mol/atom in single
> > precision gmx mdrun.
> > (
> >
> http://www.gromacs.org/Documentation/Cut-off_schemes#How_to_use_the_Verlet_scheme
> > )
> >
> > If I use double precision gmx mdrun and constraint h-bonds, can I expect
> > 1.135 rlist to give a verlet-buffer-drift as small as the one in the
> table?
> >
> > Thank you again.
> >
> --
> 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.
>