Premium
Replicated data and domain decomposition molecular dynamics techniques for simulation of anisotropic potentials
Author(s) -
Wilson Mark R.,
Allen Michael P.,
Warren Mark A.,
Sauron Alain,
Smith William
Publication year - 1997
Publication title -
journal of computational chemistry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.907
H-Index - 188
eISSN - 1096-987X
pISSN - 0192-8651
DOI - 10.1002/(sici)1096-987x(199703)18:4<478::aid-jcc3>3.0.co;2-q
Subject(s) - domain decomposition methods , molecular dynamics , intel ipsc , computer science , hypercube , domain (mathematical analysis) , decomposition , context (archaeology) , parallel computing , computational science , computational chemistry , chemistry , physics , mathematics , mathematical analysis , organic chemistry , finite element method , biology , thermodynamics , paleontology
The implementation of parallel molecular dynamics techniques is discussed in the context of the simulation of single‐site anisotropic potentials. We describe the use of both replicated data and domain decomposition approaches to molecular dynamics and present results for systems of up to 65536 Gay‐Berne molecules on a range of parallel computers (Transtech i860/XP Paramid, Intel iPSC/860 Hypercube, Cray T3D). We find that excellent parallel speed‐ups are possible for both techniques, with the domain decomposition method found to be the most efficient for the largest systems studied. © 1997 by John Wiley & Sons, Inc.