Re: length of an ABF simulation

From: Chris Chipot (
Date: Sat Nov 03 2007 - 10:44:25 CDT


estimating a priori what the length of an ABF run ought to be is not
trivial. It is strongly system-dependent and will be modulated quite
significantly if "xi" is markedly coupled to slow-relaxing degrees of

There are a number of criteria that you should follow to ascertain
that the ABF calculation has converged:

- make sure that all the bins defined between XiMin and XiMax are
   visited uniformly;

- make sure that the force distribution in each bin are well defined
   and do not change upon additional sampling;

- as a corollary, make sure that with additional sampling, the free
   energy profile does not change.

If "xi" is strongly coupled to slow-relaxing degrees of freedom, you
may want to increase the threshold beyond which the adaptive biasing
force is applied to your system. The idea behind this is to ensure
that the force distribution is well resolved and "correct" prior to
applying the biasing force. This would otherwise result in quasi
non-equilibrium effects.

Chris Chipot

Vlad Cojocaru a écrit :
> Dea NAMD users,
> I am doing some tests with the ABF method and I would like to ask you
> whether there is any way to estimate the length of an ABF simulation
> that is necessary to span the reaction coordinate values from XiMin to
> XiMax in dxi bins ? In other words, having one set of parameters is it
> possible to calculate how long the simulation needs to be ?
> Thanks
> Best wishes
> vlad


Chris Chipot, Ph.D.
Equipe de dynamique des assemblages membranaires
Unité mixte de recherche CNRS/UHP No 7565
Nancy Université Phone: +33 (0)3-83-68-40-97
B.P. 239 Fax: +33 (0)3-83-68-43-87
54506 Vandœuvre-lès-Nancy Cedex


     Science without management is worse than management without science

                                                        N. G. van Kampen

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:45:28 CST