From: Jérôme Hénin (jerome.henin_at_ibpc.fr)
Date: Tue Oct 17 2017 - 08:23:44 CDT
I think this is the explanation. subtractAppliedForce does not entirely
recover the old behavior, as the forces are subtracted at the colvar level
and not at the atom level. In your case, subtractAppliedForce should not
make a difference if you have only one active bias.
I would caution you against trusting the results though: the calculation is
still formally not correct in the case you describe. It may seem correct
due to cancellation of errors, because it is incorrect in two different
ways at the same time.
If it is indeed quantitatively correct, then there is something at play
that I don't understand yet.
On 17 October 2017 at 05:23, yjcoshc <yjcoshc_at_gmail.com> wrote:
> Hi Giacomo,
> I have tried both subtractAppliedForce on and off, and both options give
> me the same results, so I think it may not due to the total force changes.
> 在 2017年10月17日 10:43, Giacomo Fiorin 写道:
> Does this apply to that simulation?
> On Mon, Oct 16, 2017 at 9:20 PM, yjcoshc <yjcoshc_at_gmail.com> wrote:
>> I tested ABF on the cyclohexane model in this paper (
>> http://dx.doi.org/10.1021/acs.jctc.6b00447). The three dihedrals defined
>> in colvars component appeared to be not orthogonal, but
>> NAMD_Git-2017-10-09_Source with latest colvars code gave me correct results
>> while NAMD 2.11 didn't. Did ABF orthogonality or something change in
> Giacomo Fiorin
> Associate Professor of Research, Temple University, Philadelphia, PA
> Contractor, National Institutes of Health, Bethesda, MD
This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:21:43 CST