[gmx-users] Free energy - FEP

Hannes Loeffler Hannes.Loeffler at stfc.ac.uk
Mon Oct 17 20:57:11 CEST 2016


On Mon, 17 Oct 2016 18:06:01 +0200
Alex <alexanderwien2k at gmail.com> wrote:

> What I have already done in Free energy simulation was to use the
> output configuration of the last lambada windows as input
> configuration (lambada_n.gro) for the new lambada windows and so on,
> which this make simulation so time demanding as one lambada should
> start after the other one. However, I was wondering if it is
> possible(Meaningful) to simulate each windows associated with each
> lambda in FEP, completely independent to each other, so that all the
> windows could start simultaneously and then much more time would be
> saved?

Principally, each window can be considered independent of all others.
The practical problem will be that you can make sure that each window is
equilibrated adequately.  When you know both end states you may be
able to make an educated guess if the transformation will go smoothly
and you can run from the same starting state. In case you do not know
the end state, you may have to consider that that state is very
different from the initial state so "dragging" the transformation along
lambda does make some sense (but which may also fail if the end state is
totally different). In any case, try to avoid "growing" atoms/molecules
if you can or otherwise monitor carefully what is happening.  A good
test is always to run in both directions and see if the result is within
statistical bounds.


More information about the gromacs.org_gmx-users mailing list