[gmx-users] Free energy sampling using G_bar

Justin A. Lemkul jalemkul at vt.edu
Thu Sep 15 17:21:35 CEST 2011

Fabian Casteblanco wrote:
> Hello all,
> I am finished running a free energy calculation using g_bar and i
> followed Justin Lemkuls tutorial and I am in the process of analyzing
> inorder to determine if I had adequate sampling.  I have read the
> 'BAR' paper by Bennett but there are still some concerns whether I
> have enough sampling (see attached).   I see in the tutorial that the
> sampling goes up to very large sample numbers, some greater than
> 30,000 where mine only make it to about 300.  This 'samples' refers to

Sampling in the tutorial is excessive; I believe I state that in discussing the 
free energy output settings.

> how many times a certain energy level was experienced?  Does this mean

Yes.  The output histograms show the distribution of energy values in 
neighboring windows.

> my system should be left for longer times?  My lines don't line as as
> closely as the Methane Decoupling tutorial do.  Also, for other

The histograms look fine to me.  Lots of overlap.

> solvents that I used, some are giving me a warning for violating the
> 2nd law of thermodynamics and i noticed its because on one of the
> output lines, the entropy decreases a small bit, and some are giving
> me a free energy with a std deviation of at most 1.7 kJ/mol.   Does
> this all mean my sampling is not enough?

Perhaps in this instance, yes, but not all of the sampling.  You may have to 
sample a bit more in the window that's causing problems, but likely not all of 
them.  I don't know whether uneven sampling (i.e. more points in one window 
relative to others) is a problem for BAR.


> Thanks for your help.
> --
> Best regards,
> Fabian F. Casteblanco
> Rutgers University --
> Chemical Engineering PhD Student
> C: +908 917 0723
> E:  fabian.casteblanco at gmail.com


Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080


More information about the gromacs.org_gmx-users mailing list