That's correct about the reference--however, the new states ARE generated
by rescaling the volume. The box dimensions change, and the centers of
mass of each molecule move along with it. mdgx and pmemd work differently
in this respect--pmemd rescales centers of mass and thereafter treats every
molecule as a rigid body, but mdgx treats all molecules as stretchable
along with the box, insofar as the molecules themselves are flexible, again
to my recollection. The particle count is different in each case--mdgx
counts all rigid molecules (i.e. waters), all rigidly constraints groups
(i.e. SHAKE-constrained bonds), and all other atoms in the system when it
tallies up N. pmemd looks solely at the number of separate molecules, to
go hand-in-hand with its molecular virial convention.
Dave
On Wed, Nov 15, 2017 at 3:13 AM, filip <filip.persson.bpc.lu.se> wrote:
> Thank you David for providing the general protocol running in the MC
> barostat. From the code I conclude that the new states in the Markov
> chain are generated by displacing the molecules, rather than changing
> the volume, and then rescale the potential associated with the move. If
> I understand you correctly, the reference that I am after, and the one
> you based your code on, is the (quite general) protocol for the MC
> barostat outlined in Allen & Tildesley? If not, please provide the
> relevant reference that I could cite.
>
> Thanks Niel for weighing!
>
> Best,
>
> Filip
>
>
>
> On 2017-11-14 20:18, David Cerutti wrote:
> > I'm kind of the person who introduced the MC barostat into Amber, via the
> > mdgx code. Jason Swails and Scott Legrand ported it into pmemd. The
> > reference is a good one; there are also discussions about it in
> > Computational Simulation of Liquids by Allen and Tildesley if I recall
> > correctly.
> >
> > In order to better understand the MC barostat, it's useful to consider
> the
> > case of an ideal gas, and how the barostat does indeed correctly
> reproduce
> > the intended pressure / volume relationship despite the naïve perception
> > that a volume move on an ideal gas would not produce a change in the
> > potential energy of the particles (which is purely kinetic). It's all
> > about the effect of the work term PdV and the probability of making the
> > move (see Barostats.c, line 832), which includes the number of particles
> N
> > and the log of the volume change, on the metropolis acceptance criterion.
> > That's the detailed balance relationship and the means by which PV = (((
> N
> > ))) RT.
> >
> > Dave
> >
> >
> > On Tue, Nov 14, 2017 at 1:17 PM, Niel Henriksen <shireham.gmail.com>
> wrote:
> >
> >> Hi Filip,
> >>
> >> I did not implement the MC barostat in AMBER, but this paper should
> provide
> >> useful background:
> >> https://doi.org/10.1016/j.cplett.2003.12.039
> >>
> >> The relevant AMBER code is
> >> here: $AMBERHOME/AmberTools/src/sander/barostats.F90
> >>
> >> --Niel
> >>
> >> On Tue, Nov 14, 2017 at 3:09 AM, filip <filip.persson.bpc.lu.se> wrote:
> >>
> >>> Dear AMBER developers,
> >>>
> >>> I would like to see a reference for the Monte Carlo barostat
> implemented
> >>> in AMBER 16 in order to understand exactly how it works. The manual is
> >>> lacking in detail here and is not of very much help unfortunately.
> >>>
> >>> Best regards,
> >>> Filip Persson
> >>>
> >>> PhD student
> >>> Department of Biophysical Chemistry
> >>> Lund University
> >>> Sweden
> >>>
> >>>
> >>> _______________________________________________
> >>> AMBER mailing list
> >>> AMBER.ambermd.org
> >>> http://lists.ambermd.org/mailman/listinfo/amber
> >>>
> >> _______________________________________________
> >> AMBER mailing list
> >> AMBER.ambermd.org
> >> http://lists.ambermd.org/mailman/listinfo/amber
> >>
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Nov 15 2017 - 00:30:03 PST