[gmx-users] free energy coupling/decoupling equivalence

[Berk Hess]

Hi,

If you are talking about equilibrium methods, for instance free energy intergration
or Bennett's acceptance ratio, there is no directionality.
You might start a simulations at a lamba point with the end configuration of the previous lambda,
which could introduce some directionality, but then you should equilibrate long enough
to see no effect of this.

If you are doing non-equilibrium methods, than I would say that it is common sense
that you should start from the higher entropy state, since starting from the lower
entropy state increase the probability that you miss a significant part of the phase
space of the other state.

Berk

> Date: Thu, 12 Mar 2009 22:18:25 -0400
> From: chris.neale at utoronto.ca
> To: gmx-users at gromacs.org
> Subject: [gmx-users] free energy coupling/decoupling equivalence
> 
> Hello,
> 
> I have just read an interesting paper suggesting that the reference state in
> free energy decoupling/coupling simulations should be the one with larger
> entropy. Lu and Kofke "Accuracy of free-energy perturbation calculations in
> molecular simulation. I. Modeling" JCP V. 114, N. 17 pp. 7303-7311 (2001).
> 
> I have seen some mention that:
>    <begin quote>
>      Ordinarily, one would expect that "forward" and "reverse" only are
>      meaningful when you're (a) doing nonequilibrium simulations, or (b)
>      beginning simulations at lambda values from the ending configurations
>      at preceding lambda values.
>    <end quote>
> http://www.gromacs.org/pipermail/gmx-users/2008-February/032430.html
> 
> And now I am confused, since I would have thought that this was an important
> "forward" vs. "reverse" difference, although I may have my terminology  
> wrong here.
> 
> To put my questions succinctly:
> 
> 1. In the gromacs implementation, is there any important difference
>     between coupling and decoupling? I avoid the terms "forwrad" and
>     "reverse" in order to avoid non-equilibrium or slow-growth connotations.
> 
> 2. Does soft-core work properly for both LJ coupling and LJ decoupling?
>     I am not clear how mdrun knows whether to soft-core lambda or 1-lambda
>     since it is near the decoupled state that really needs the soft-core,
>     and this could be at lambda or 1-lambda.
> 
> Background information:
> I am doing coulomb and LJ separately; LJ only with soft-core; coulomb  
> first in the decoupling. Everything is working fine for me in the  
> decoupling department, but I am now considering switching to coupling  
> as opposed to decoupling based on this Lu 2001 paper and am interested  
> if anybody has some comments.
> 
> Thank you,
> Chris.
> 
> 
> 
> 
> 
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the 
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today it's FREE!
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20090313/59e8e218/attachment.html>