AMBER Archive (2009)

Subject: Re: [AMBER] Partial interactions

From: steinbrt_at_rci.rutgers.edu
Date: Tue Apr 21 2009 - 10:17:55 CDT

Hi Ignacio,

> and try to modify it. Is there any document explaining the logic of the
> code, what each file does, how the lists of atoms are managed, etc? Or

<insert malicious laughter here> no, there isnt, as far as I know an Amber
programmers manual doesnt exist. It can be daunting for a beginner to
modify the code, it may even be easier to switch to something like Tinker
for what you have in mind. Amber is written to be highly efficient and
parallel which makes the code fairly complex. It being a group effort of
many contributors with different programming styles doesn't help
readability either (but it adds to the fun of working on it :-) )

If you stick to Amber, you could have a look at the call to force() in
runmd, in the $AMBERHOME/src/sander directory. This is where all the
interactions are computed. Following the code in force.f will get you to
the points where all the different interaction types are computed and I
guess you would have to add exceptions to each of them. A problem would be
the Ewald routines, as the interaction in the reciprocal part isn't really
pairwise decomposable.

Overall I would rate what you have in mind to be a fairly complex project
to get right, but maybe Im overly pessimistic here...

Kind Regards,

Thomas

Dr. Thomas Steinbrecher
BioMaps Institute
Rutgers University
610 Taylor Rd.
Piscataway, NJ 08854

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber