[gmx-users] Error estimation from FEP

Ragnarok sdf fabracht1 at gmail.com
Thu Sep 10 14:21:52 CEST 2009


I have done FEP simulations and am trying to find a way to report the error
related to the values of relative delta G found. I have simulated 20 windows
each with an interval of lambda of 0.05. Then used -rerun option from mdrun
to perform the same simulation with a lambda value of (lambda+0.05) and
(lambda-0.05). G_energy gives out dF as being an average of the exponential
value of the difference between the potential energy of my state A and my
state B in a logarithmic scale, right?
Well I found a lecture given by Berk Hess on Gromacs Workshop in 2007, and
there is says that the error estimate could be given by g_analyze -ee. Well,
Would this error estimate be over the final value of dF (= -kT ln <
exp(-(EB-EA)/kT) >A) over several blocks of my simulation? Being so, I would
need to organize the values in a single column so as to look like an xvg
input file for g_analyze?
Being the final delta delta G a sum of all different lambda intervals, would
the error estimate of my final sum be the sum of errors for each interval?
Thank you in advance
Fabrício Bracht
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20090910/71885aac/attachment.html>


More information about the gromacs.org_gmx-users mailing list