From chemistry-request@ccl.net Fri Sep 13 18:48:14 1991
Date: Fri, 13 Sep 91 17:41:05 CDT
From: heller@lisboa.ks.uiuc.edu (Helmut Heller)
To: chem@lisboa.ks.uiuc.edu
Subject: RE: MD neighbor lists
Status: R

As a reply to John's question I would like to point out some relevant articles:

@article{SKEE91,
author={Robert D. Skeel},
title={Macromolecular Dynamics on a Shared-Memory Multiprocessor},
journal=JCC,
volume=12,
number=2,
pages={175--179},
month=jan,
year=1991
}
Skeel discusses various methods to generate neighbor lists (on parallel computers)  
and concludes that grid cells plus linked lists are best.

I would like to suggest NOT to use a cut-off, but instead use the Fast Multipole  
Methods or variable time step methods (distance classes). The latter has been  
implemented by Helmut Grubm"uller and myself in a parallel MD program for a  
Transputer system and is described in:
@article{GRUB91,
author={Helmut Grubm\"uller and Helmut Heller and Andreas Windemuth and Klaus  
Schulten},
title={Generalized Verlet Algorithm for Efficient Molecular Dynamics Simulations  
with Long-Range Interactions},
journal=MS,
year=1991,
volume=6,
pages={121--142},
number={1--3}
}

while the computer system and MD program used is described in
@article{HELL90,
  author={Helmut Heller and Helmut Grubm\"uller and Klaus Schulten},
  title={Molecular Simulation on a Parallel Computer},
  journal=MS,
  volume=5,
  year=1990,
  pages={133--165}
}

@string{MS = "Molecular Simulation"}
@string{JCC = "J.\ Comp.\ Chem."}

Cheers, Helmut  (W9/DH0MAD) _________________________________________________
heller@lisboa.ks.uiuc.edu   "Knowledge must be gathered and cannot be given" 

u27013@ncsagate.bitnet                   ZEN, one of BLAKES7                

Phone: (217)244-1586, FAX: (217)244-2909 ------------------------------------
Helmut Heller, University of Illinois at Urbana-Champaign, Beckman Institute
Theoretical Biophysics Group, Transputer Lab, Room 3151, MC 251
405 N. Mathews Ave., Urbana, IL 61801, U.S.A.