Re: FEP - one long simulation vs many short ones

From: Jérôme Hénin (
Date: Sat Jul 02 2005 - 18:45:11 CDT

Dear Floris,

That's an very interesting question, because as you mention, in the case
of a large distributed computing infrastructure with poor or average
communication performance, running many independent FEP windows is
obviously a very efficient approach. So how necessary is it to follow
the "traditional", sequential protocol?

>I have a question about my approach to running FEP

>I wanted to ask if there are any dangers in running
>FEP like this?
Formally, the theory for the staged version of FEP doesn't put any
requirement on how the different stages are sampled. So in the limit of
infinite sampling time, this is a non-issue (as most of our usual
problems). It goes without saying that, in practice, getting the FEP
average to converge *is* an issue.

>Is this a common approach?
To the best of my knowledge, it isn't. I know at least one guy who has
done it recently, though, so he may have valuable remarks about that.
Phil, I know you are reading this thread, can you comment on this?

> I guess I
>risk problems in the equilibration from the lambda 0.5
>starting point to the test value. However perhaps I
>gain something by cutting out any general 'drift' that
>would inevitably arise over a long simulation, which
>in turn reduces or bypasses the issue of hysteresis
>associated with a 'directional' simulation.
Well, the canonical point of view would probably imply that the "drift"
means that your system is moving to regions of phase space
well-characterized by the Hamiltonian with the current value of lambda.
In other words, the system should be following the Hamiltonian.
You raise a good point there, however: what part of the calculated
hysteresis is due to incomplete sampling or poor ensemble overlap, and
what part results from some slow degrees of freedom having drifted
between the forward and reverse calculations? I don't think this
distinction is clearly made in most papers presenting FEP results. I
have to say, however, that if you come to fear that your system could
drift in a non-physical/non-desired way if you run a longer simulation,
then you somehow distrust your setup or forcefield/simulation
conditions. Then you might want to consider carefully whether it is
worth carrying out free energy calculations at all. I mean no offense
there, that's a question I've already asked myself in the past.

As I see it, your question boils down to this: how fast does the system
react to a change in lambda? So the answer is bound to be
system-dependent. There should be ways to quantify this relaxation time,
however. Perhaps we should have a close look at these papers of the 90's
on Hamiltonian lag again - a part of the answer might already be in there.


This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:40:55 CST