Parallel implementation of a Monte Carlo molecular simulation program

• Type: Article in International Scientific Journal
• Date: 2000

Authors

Carvalho, AP (Author) Other; Gomes, JANF (Author) FCUP; Cordeiro, MNDS (Author) FCUP

Journal

Title: Journal of Chemical Information and Computer Sciences

Vol. 40

Pages: 588-592

ISSN: 0095-2338

Publisher: American Chemical Society

Indexing

ISI Web of Knowledge - 6 Citations

Scopus - 6 Citations

Scientific classification

FOS: Natural sciences > Chemical sciences

Other information

Authenticus ID: P-001-06F

DOI: 10.1021/ci990101k

Abstract (EN): Molecular simulation methods such as molecular dynamics and Monte Carlo are fundamental for the theoretical calculation of macroscopic and microscopic properties of chemical and biochemical systems. These methods often rely on heavy computations, and one sometimes feels the need to run them in powerful massively parallel machines. For moderate problem sizes, however, a not so powerful and less expensive solution based on a network of workstations may be quite satisfactory. In the present work, the strategy adopted in the development of a parallel version is outlined, using the message passing model, of a molecular simulation code to be used in a network of workstations. This parallel code is the adaptation of an older sequential code using the Metropolis Monte Carlo method. In this case, the message passing interface was used as the interprocess communications library, although the code could be easily adapted for other message passing systems such as the parallel virtual machine. For simple systems it is shown that speedups of 2 can be achieved for four processes with this cheap solution. For bigger and more complex simulated systems, even better speedups might be obtained, which indicates that the presented approach is appropriate for the efficient use of a network of workstations in parallel processing.

Language: English

Type (Professor's evaluation): Scientific

No. of pages: 5