From: JC Gumbart (gumbart_at_ks.uiuc.edu)
Date: Wed Jul 27 2011 - 23:27:23 CDT
It didn't appear like the reference atoms were changing structure much, but I'll have to look at it more quantitatively. In fact, I'm using a lot of colvars precisely to limit structural changes during FEP.
On Jul 27, 2011, at 9:20 PM, Giacomo Fiorin wrote:
> Hi JC, it means that the function that computes the rotation matrix can't find an optimal rotation between the current and the reference coordinates, because the current structure has changed dramatically. Typically this occurs when a roto-reflection brings the current coordinates closer to the reference than a pure rotation does. In that situation, the rotation is continuous in the coordinates, but its gradients (and the forces) are not.
>
> Does your set of reference atoms change structure much?
>
> To avoid that, you could try using a narrower selection of atoms, on a section that doesn't change structure.
>
> Giacomo
>
> On Wed, Jul 27, 2011 at 9:41 PM, JC Gumbart <gumbart_at_ks.uiuc.edu> wrote:
> I am trying to run an FEP simulation under the influence of a number of colvars, including RMSD, orientation, angles, etc. However, it crashes quite frequently due to atoms moving too fast, immediately preceded by a "discontinuous rotation" warning. Although I looked at the source code, it's still unclear to me what is specifically happening in my simulation to cause it. Does anyone have suggestions for ways to avoid it? It's becoming quite frustrating.
>
> Thanks!
>
>
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:57:31 CST