Re: [AMBER] Add extra force in sander.MPI

From: case <case.biomaps.rutgers.edu>
Date: Sun, 2 Oct 2011 13:31:58 -0400

On Sun, Oct 02, 2011, Chicago Ji wrote:
>
> I'll creat another array ext_force(natom*3), and broadcast the whole
> information of ext_force(*) to all threads.
> And then add ext_force(*) to original f(*) .

This sounds more complicated that is needed. In sander, every thread
knows the coordinates of all atoms, and any thread can calculate
the energy and forces for any part of the energy function, adding
contributions in the proper place to [its own version of] f(*). You,
as a programmer, do not have to worry about the bookkeeping that is
involved in getting those forces transferred to the proper place for
updating the coordinates in runmd.

(Basically, sander is designed to make it as easy as possible to add extra
terms to the energy. The price that is paid is that more communication is
needed, compared to a code like pmemd.)

....dac


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun Oct 02 2011 - 11:00:21 PDT
Custom Search