Re: AMBER: Amber 9 TI Details

From: Thomas Steinbrecher <>
Date: Tue, 27 Mar 2007 14:58:55 -0700 (PDT)

Hi Lachele,

the way amber9 does TI under the multisander framework is that one group
runs using the V0 potential from the first group's prmtop and the other
uses the V1 potential of prmtop file #2. At any given MD step, after all
forces have been calculated, they are mixed according to whatever
combination rule you decided to use (in routine mix_frcti which you'll
find in thermo_int.f) and rebroadcast so that both processes
have identical force arrays. At this point DV/DL is calculated from the two
different potential energies that processes 0 and 1 calculated. After
that the energies are also mixed so that they correspond to the combined
potential function at the lambda value you chose. Then both processes
propagate the coordinates as usual and, unless something goes very wrong
end up having the same coords for the start of the next MD step. Your
group 0 and group 1 output files should thus always be identical apart
from trivial stuff like timings and output filenames.

In amber 9 it is necessary that both processes have the same number of
atoms, so you have to replace vanishing ones with dummy atoms. At the
moment I am developing a new approach to this so that in amber10 you
should be able to just use prmtop files for the initial and end state
without putting dummy atoms into them.

I hope this answers your first question, as for the second one, both
processes use the masses from prmtop file #1 so it does not matter what
mass you set for your dummy atoms (or any atom in prmtop2, they are just
ignored). I have always accepted the quoted sentence from the manual in
good faith, but you're right, I'd like to have a reference for that too.

Kind Regards,


Dr. Thomas Steinbrecher
The Scripps Research Institute
10550 N. Torrey Pines Rd.
San Diego CA 92037, USA
The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Wed Mar 28 2007 - 06:07:47 PDT
Custom Search