From: Jérôme Hénin (jerome.henin_at_ibpc.fr)
Date: Tue Feb 24 2015 - 16:18:21 CST
The only requirement here is that the sampling you get from each window is
equilibrated at the corresponding value of lambda. If you start from a
nearby value of lambda the equilibration time required will be shorter.
On 24 February 2015 at 09:59, Jian Dai <djpittdj_at_gmail.com> wrote:
> Dear all:
> I have questions about simulation scheme in FEP.
> First, for example, if I divided the simulation into 10 windows for lambda
> from 0 to 1, so that they are 0 - 0.1 (window 0), 0.1 - 0.2 (window 1), ...
> 0.9 - 1.0 (window 9), should I use the final coordinates of window 0 to
> conduct simulation on window 1, and so on? Or in principle I could use the
> same starting coordinates for all 10 windows? I guess it should be the
> former scheme.
> Second, if I later decide to add some intermediate windows, for example, I
> want to add a window corresponding to lambda 0.05 - 0.1, is this
> feasible?Or should I plan how the lambda should change before I start
> Thank you in advance.
This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:20:55 CST