From: Jérôme Hénin (jerome.henin_at_ibpc.fr)
Date: Tue Jun 11 2013 - 07:09:06 CDT
This problem can be annoying, it is caused when the discretization of the mean force used by ABF breaks down due to a singularity in the PMF. You can take two kinds of measures to limit this:
1) avoid biasing the wall region: maybe not by restraining the coordinate, but rather by changing the grid limit (lower/upperBoundary) so that the very steep region gets off-grid.
2) refine the grid. If you have several windows, this could be done locally, close to the hard wall.
3) one option I've been pondering is to add a cap on the biasing force applied, which could work in your case. That could be implemented soon, in which case we'll let you know about it.
----- Original Message -----
> I'm doing simple adsorption energy calculations using ABF onto a hard
> surface (one coordinate). Everything works fine, but often the
> simulation gets stuck trying to resolve the potential very near the
> surface, which is very large but not interesting or important.
> I've tried restraining the colvar to a minimum of several angstroms
> from the surface which helps, but the offset is different for each
> molecule (different adsorption distances), and if it's too large the
> minimum in the resulting PMF is missed.
> Does anyone have a good suggestion for doing ABF calculations when
> there is a hard wall present?
This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:21:17 CST