Re: FEP: delta_G, memory of Hamiltonians corresponding to last G value?

From: Jerome Henin (
Date: Thu Oct 09 2008 - 17:49:05 CDT

Dear Sebastian,

NAMD does not remember the deltaH values because they cannot be reused
for the next window as they are sampled from the wrong ensemble. Let
me explain this : if you go from A to B to C, then deltaG_AB is
computed based on (E_B - E_A) values, sampled in state A, whereas
deltaG_AB is computed based on (E_C - E_B) values sampled in state B.
So you can't use the distribution of E_B obtained in state A for the
second window (B->C).

As a side note: some greatly improved FEP code has been making its way
into CVS lately. If you are not afraid of compiling NAMD, you might
want to have a look.


On Thu, Oct 9, 2008 at 3:10 PM, Sebastian Stolzenberg
<> wrote:
> Dear All,
> in NAMD-FEP, say, I execute the command
> runFEPlist {.001 .01 .1} $numSteps
> After having calculated delta_G(.001)=G(.01)-G(.001) with exponential
> averages of Hamiltonians, will NAMD "memorize" the Hamiltonians
> corresponding to G(.01) when next calculating delta_G(.01)=G(.1)-G(.01)? Or
> will it re-measure the Hamiltonians corresponding to G(.01) and then proceed
> to calculate G(.1) and delta_G(.01)?
> Thank you,
> Best,
> Sebastian

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:49:57 CST