Re: Using TCLforces to rescale force(s) on particular atom(s) --- loadtotalforces, addforce and previous/current timestep issue

From: Giacomo Fiorin (giacomo.fiorin_at_gmail.com)
Date: Thu Jan 19 2012 - 11:56:00 CST

Hello Gungor, I'm afraid not, the forces are by design those at the
previous time step, because collecting the forces from the current time
step before it is completed could slow down NAMD significantly. The idea
is that while one node executes TCL forces, some of the slower nodes are
still completing e.g. the vdW interactions.

Maybe, since you're working with few atoms, you can use an integrator of
higher order than the velocity Verlet to *extrapolate* the force at the
current time step based on those at the previous time steps? Obviously,
you should factor in the forces you're adding as well.

Giacomo

On Thu, Jan 19, 2012 at 12:44 PM, Gungor Ozer <gungor_at_bu.edu> wrote:

> Dear all,
>
> I have been trying to adapt a TCL script that would use TCLforces to
> rescale force on specified atoms. The idea is simple:
>
> 1) Calculate current force on a particular atom, f ---
> (loadtotalforces)
> 2) Rescale it using some factor, x.
> 3) Add force, (x-1)*f --- (addforce)
>
> I was planning to do this using the built-in loadtotalforces and addforce
> commands. However, there is one issue: loadtotalforces reads the total
> forces on each requested atom in the "previous timestep" but addforce adds
> the force at the "current time step".
>
> Is there anyway to get around this? Any input will be helpful and greatly
> appreciated.
>
> Thanks in advance,
> Gungor
>

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:21:08 CST