From chemistry-request@ccl.net Tue Feb 25 01:27:05 1992
Date:    Tue, 25 Feb 1992 1:09:19 -0500 (EST)
From: GIVEN@sbchm1.chem.sunysb.edu
Subject: Large-Scale MD for Coulomb Forces
To: chemistry@ccl.net
Status: R

Dear Friends,
  I am re-writing an MD code for a model system of charged spherically 
symmetric ions having pairwise interactions consisting of smooth
short-range repulsive core terms combined with pure (i.e., unscreened) Coulomb 
interactions. 
The old code spent most of its time calculating Coulomb forces. 
Since my background is in physics and not chemistry, I wondered why the
Greengard-Rokhlin (GR) algorithm would not be the fastest way to handle the 
force calculation. (The GR algorithm is the version I am familiar with of an
intelligent multipole expansion method, essentially an optimized use of Gauss' 
law.)
Is this in fact how people handle MD for such systems, i.e., for 
systems with unscreened Coulomb interactions? Is there a version of the
Greengard-Rokhlin algorithm available anywhere that is optimized, i.e., either 
by vector routines or parallel programming? 

  I was originally led to this line of reasoning by attending the Materials
Research Society meeting in Boston and hearing brave tales of colloidal system 
simulations involving >10,000 particles. State of the art simulations of the 
restricted primitive model, including those used to explore its critical point, 
are FAR smaller. The colloidal system simulations make a virtue of the screened 
nature of the (MacMillan-Mayer level) forces. I want to make a virtue out of 
the pure 1/r**2 nature of Coulomb forces. 

  If this is an "easy" question, I would be grateful for a recent footnote
to an article describing an efficient way to handle the Coulomb force
calculation. 

  I will share any insights that are not well known. Thanks for your
consideration

                                  James A. Given
                                  Computational Chemistry Group
                                  Dept. of Chemistry SUNY
                                  Stony Brook, NY
                                  in%"sbchm1.chem.sunysb.edu"






From chemistry-request@ccl.net Tue Feb 25 01:53:23 1992
Date:    Tue, 25 Feb 1992 1:36:39 -0500 (EST)
From: GIVEN@sbchm1.chem.sunysb.edu
Subject: Large-Scale MD for Coulomb Forces
To: chemistry@ccl.net
Status: R

Dear Friends,
  I should have been more explicit. My e-mail ID for discussion
of Coulomb force vector and parallel codes is:
                in%"given@sbchm1.chem.sunysb.edu"
Thanks again. I am
                                  Yours Truly
                                  James A. Given

  

From jkl@ccl.net Tue Feb 25 02:07:57 1992
To: chemistry@ccl.net
Subject: Subscription list
Date: Tue, 25 Feb 92 02:07:48 EST
From: jkl@ccl.net
Status: R

I must have said something different than I wanted to. I WILL NOT DISTRIBUTE
the list of Computational Chemistry subscribers. I will only send you a SINGLE
address of a SPECIFIC person if you request it. So please, before you send
your request to chemistry@ccl.net, try me (i.e., jkl@ccl.net) first.
Jan Labanowski
jkl@ccl.net




From chemistry-request@ccl.net Tue Feb 25 15:53:11 1992
Date:    Tue, 25 Feb 1992 14:43:39 -0500 (EST)
From: GIVEN@sbchm1.chem.sunysb.edu
Subject: Large-Scale MD for Coulomb Forces
To: chemistry@ccl.net
Status: R

Dear Friends,
  Several people that responded to my request for an implementation of the 
Greengard-Rokhlin algorithm expressed great interest in the possibility of 
a FAST algorithm for doing the Coulomb calculations in MD simulation of 
atomic systems. In response to two requests for more detailed descriptions of 
the algorithm, I note the following:

1. "A Fast Algorithm for Particle Simulations", L. Greengard and V. Rokhlin, J. 
Comp. Physics, {\bf 73}, 325 (1987)
2. "Vectorization of a Tree Code", J. Makino, J. Comp. Physics, {\bf 87}, 148 
(1990)
3. J. Barnes and P. Hut, Nature, {\bf 324}, 446 (1986)

  The #2 article above discusses the possibility of an efficient vector code 
for this task. I still don't know of any implementation of this kind (but
promise to pass that knowledge on if I receive it.)

                                         Best Wishes,
                                         Jim Given
                                         in%"given@sbchm1.chem.sunysb.edu"


From chemistry-request@ccl.net Tue Feb 25 18:52:59 1992
Date: Tue, 25 Feb 1992 23:26:23 GMT
From: hughc@extro.ucc.su.OZ.AU (Hugh Capper)
To: chemistry@ccl.net
Subject: Gaussian set of programs
Status: R

Dear netters,
I am enquiring as to which platforms the Gaussian set of programs has been ported.  Can anyone supply me with a list of compatible platforms?

                                   Regards
                                   Hugh Capper