[gmx-developers] Elaboration of the error message "Replica exchange is only supported by dynamical simulations"

[Michael R Shirts]

> 1. Would a MC simulation, where moves are randomly chosen and accepted according to Metropolis criterion, be considered random integration?
> 2. Is there something theoretical that disqualifies random integrators from parallel tempering, or are they simply not supported by GROMACS at this time?
There is no issue with performing parallel tempering with Metropolis Monte Carlo.  Parallel tempering simply allows exchange between systems that are sampling from equilibrium distributions.  It can be used with any such simulations, no matter the type.  Different types of sampling might require different implementations of the same math, though. I think “dynamical” simulation just means one simulation is carrying out sampling, not doing something else.  Perhaps the error could be a bit clearer.

Best,
~~~~~~~~~~~~~~~~
Michael Shirts
Associate Professor
michael.shirts at colorado.edu<mailto:michael.shirts at colorado.edu>
http://www.colorado.edu/lab/shirtsgroup/
Phone: (303) 735-7860
Office: JSCBB C123
Department of Chemical and Biological Engineering
University of Colorado Boulder


From: <gromacs.org_gmx-developers-bounces at maillist.sys.kth.se> on behalf of Gregory Schwing <go2432 at wayne.edu>
Reply-To: "gmx-developers at gromacs.org" <gmx-developers at gromacs.org>
Date: Thursday, June 7, 2018 at 9:11 AM
To: "gromacs.org_gmx-developers at maillist.sys.kth.se" <gromacs.org_gmx-developers at maillist.sys.kth.se>
Subject: [gmx-developers] Elaboration of the error message "Replica exchange is only supported by dynamical simulations"



Dear GMX-Dev Mailing List:



I am interested in adding parallel tempering to a Monte Carlo simulation engine known as GOMC.  Naturally, I am using the GROMACS implementation as a guide.  However, GOMC is not an MD engine, so I was concerned when I read the error message noted in the subject of the email, found in repl_ex.cpp.  My question is twofold:



  1.  Would a MC simulation, where moves are randomly chosen and accepted according to Metropolis criterion, be considered random integration?
  2.  Is there something theoretical that disqualifies random integrators from parallel tempering, or are they simply not supported by GROMACS at this time?



Sincerely



Greg
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-developers/attachments/20180607/f83cafa9/attachment-0003.html>