From: Mark M Huntress (
Date: Mon Apr 04 2011 - 15:04:17 CDT

 I am not sure what you mean by rescaling the positions. and if I understand what you mean by "The velocities should certainly be independent of the positions and
translational invariant," I think I might have a problem. When I recenter, there are several solvent molecules whose x y or z coordinates are translated in order the shift the solvent box so that it surrounds the solute molecule. Let us take for example one water whose x coordinates are shifted. Before recentering, its velocity is in the positive x direction --> and it is going toward the solute molecule:

water --> solute

After recentering, the x coordinates of the water are shifted so that it is on the other side of the solute. If the velocity is invariant, that would make it so that the water is now moving away from the solute:

                       solute water -->

Am I understanding this correctly? If so, does this accurately reflect how the solvent is modeled with periodic boundary conditions? Would I be making an error to start a new QMMM trajectory with these velocities after re-centering?
From: Olaf Lenz []
Sent: Monday, April 04, 2011 3:11 AM
To: Mark M Huntress
Subject: Re: vmd-l: recenter


On 04/03/2011 11:34 PM, Mark M Huntress wrote:
> I am doing a classical MD trajectory of an organic molecule in a
> solvent box, and I am sampling frames (coordinates and velocities)
> from this MM trajectory to start new QMMM trajectories. For the
> frames that I am sampling, I am re-centering the solute in the
> solvent box (using pbc wrap -centersel {resid 1} -center com
> -compound res), and my concern is whether I can still use the
> velocities from the vel.dcd file. Will the coordinates and velocities
> be incompatible for any reason after I re-center? (for instance,
> reordering of solvent atoms or the velocities of some solvent atoms
> becoming in the wrong direction relative to the molecule?) Thanks,

The velocities should certainly be independent of the positions and
translational invariant. As long as you don't rescale the positions,
everything is safe.


Dr. rer. nat. Olaf Lenz
Institut für Computerphysik, Pfaffenwaldring 27, D-70569 Stuttgart
Phone: +49-711-685-63607