Re: [AMBER] Berendsen vs. Monte Carlo

From: Brian Radak <>
Date: Fri, 25 Jul 2014 10:49:06 -0400

If the NVT density is close to the average NpT density then, for suitably
large N, most expectations (including free energies) will be statistically
indistinguishable between the two. The basis for this ensemble equivalence
is very well described in Allen and Tildesley's book, for example.

Since NpT calculations are often less efficient (although this is much less
true now with the MC barostat), NVT is often favored. I generally think of
NpT as a convenient/lazy choice, since you don't have to worry about
picking a "good" density, provided that you think the barostat will sample
the volume space well.


On Fri, Jul 25, 2014 at 8:23 AM, Parker de Waal <>

> Hi Jason,
> Thank you for the reply, I'm feeling a lot more comfortable now about the
> Monte Carlo barostat.
> Quick question however: I've noticed that a lot of the long time scale
> simulations done by the Shaw group use NPT for their production ensemble,
> however the AMBER aMD paper (LCT Pierce et al.) was able to sample similar
> space using the dihedral/total energy boosts in the canonical NVT ensemble.
> I'm wondering if I could also use the NVT ensemble for aMD simulations
> where I expect a large structural change or if I should try out NPT.
> Thanks again,
> Parker
> ________________________________________
> From: Jason Swails []
> Sent: Friday, July 25, 2014 1:20 AM
> To: AMBER Mailing List
> Subject: Re: [AMBER] Berendsen vs. Monte Carlo
> > On Jul 24, 2014, at 10:14 PM, Parker de Waal <>
> wrote:
> >
> > Hi Everyone,
> >
> > I haven't been able to find any literature, besides the manual,
> describing the implementation of the Monte Carlo barostat in AMBER14 and
> was wondering if someone could provide some insight.
> It is described in the Allen and tildesley book. There is a brief
> description in the amber 14 manual (check the mdgx chapter as well, there
> might be more information there as well).
> You can also look at the code (barostats.F90). It's actually a remarkably
> simple algorithm, so you may find the code enlightening.
> >
> > Currently I'm trying to sample the energy landscape of an apo protein
> using aMD that I suspect is very unstable, due to the geometry of the
> ligand binding pocket, and suspect will undergo a large significant change.
> Because of this I would like to use the NPT ensemble during my production
> MD instead of NVT to allow a greater range in energy fluctuation. If
> possible I would like to use ntb =2, ntp =1, barostat = 2 to save computer
> time and speed along the process, however I'm concerned that the Monte
> Carlo barostat won't be as effective as the Berendsen implementation to
> achieve this goal.
> It's possible that it may take a while for the MC barostat to find an
> effective step size. But it is actually a more rigorously correct barostat.
> And quite a bit faster for pmemd.cuda.
> HTH,
> Jason
> --
> Jason M. Swails
> BioMaPS,
> Rutgers University
> Postdoctoral Researcher
> _______________________________________________
> AMBER mailing list
> _______________________________________________
> AMBER mailing list

================================ Current Address =======================
 Brian Radak                                            :     BioMaPS
Institute for Quantitative Biology
 PhD candidate - York Research Group       :     Rutgers, The State
University of New Jersey
 University of Minnesota - Twin Cities          :     Center for
Integrative Proteomics Room 308
 Graduate Program in Chemical Physics     :     174 Frelinghuysen Road,
 Department of Chemistry                          :     Piscataway, NJ
08854-8066                                 :
AMBER mailing list
Received on Fri Jul 25 2014 - 08:00:02 PDT
Custom Search