[MMTK] Langevin dynamics using CalphaForceField

Konrad Hinsen hinsen at cnrs-orleans.fr
Fri Sep 5 14:35:55 UTC 2008


On Sep 5, 2008, at 9:18, Geoff Rollins wrote:

> I'm interested in running Langevin dynamics simulations using the  
> Calpha force field. My understanding is that this force field takes  
> every given configuration to be a potential energy minimum, so that  
> the gradients will be zero at every time step. The test simulations  
> that I've run seem to confirm this since the Calpha's of my protein  
> gradually drift away from each other in all directions.
>
> I am wondering if there is any simple way to change the default  
> behavior so that only the initial configuration would be a  
> potential energy minimum, i.e. the energy and its gradients would  
> not be zero at every step. Or, if no simple way exists, I'm  
> wondering what sorts of modifications I would need to make in order  
> to achieve this.

Unfortunately your project requires a significant modification to  
CalphaForceField, although it is not a difficult one to make. If you  
want meaningful forces from an harmonic potential, you need to  
provide both the current configuration and a reference configuration  
that defines the minimum of the energy. The current CalphaForceField  
does not have a reference configuration argument at all, so such an  
argument would have to be added. The other modifications (calculating  
the forces) would be rather trivial.

In my opinion it is not a good idea to use the initial configuration  
as a reference. After all, what does "initial" mean in a complex  
script? The first configuration ever used in the script? The first  
one for each dynamics run? Or yet something else?

Konrad.
--
---------------------------------------------------------------------
Konrad Hinsen
Centre de Biophysique Moléculaire, CNRS Orléans
Synchrotron Soleil - Division Expériences
Saint Aubin - BP 48
91192 Gif sur Yvette Cedex, France
Tel. +33-1 69 35 97 15
E-Mail: hinsen at cnrs-orleans.fr
Web: http://dirac.cnrs-orleans.fr/~hinsen/
---------------------------------------------------------------------






More information about the mmtk mailing list