[gmx-users] Isobaric - isothermal ensemble and Replica Exchange

Berk Hess gmx3 at hotmail.com
Wed Nov 16 17:48:58 CET 2005

>From: Luca Mollica <YLU at zurich.ibm.com>
>Reply-To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>To: gmx-users at gromacs.org
>Subject: [gmx-users] Isobaric - isothermal ensemble and Replica Exchange
>Date: Wed, 16 Nov 2005 15:10:07 +0100
>Dear all,
>as previously posted in the mailing list, a Replica Exchange Molecular
>Dynamics (REMD) simulation has been performed on a protein-ligand-
>water system: each replica has been simulated at the same pressure (1
>atm) and different temperatures in order to sample temperature space.
>Therefore, NPT replicas have been used, and this does NOT of course
>represent a random walk distribution, with its probability
>distribution having a decreasing exponential trend as pointed in the
>paper published on Chemical Physics Letters by Okabe et al in 2001
>(225, 435-439).
>There is a question that I have now: if in random walk like
>distribution an averaged observable is obtainable & computable, what
>about the physical significance of parameters that can be obtained
>using REMD in isobaric-isothermal ensemble ?
>Do you have any suggestions about the data treatment for getting
>relevant parameters about the system, e.g., thermodinamic quantities ?

The pV term in the exchange criterion is usually negligible compared
to the energy term, about a factor 1000 smaller.

I don't know which Gromacs version you have used, but Gromacs 3.3
has proper NPT REMD.
The extra term is:
(P1/kT1 - P2/kT2)(Vol1 - Vol2)
where P is the reference pressure and Vol the instantaneous volume.
You can determine this term and check that it is negligble.


More information about the gromacs.org_gmx-users mailing list