Scalable time-parallelization of molecular dynamics simulations in nano mechanics

Yanan Yu, Ashok Srinivasan, Namas Chandra

Research output: Chapter in Book/Report/Conference proceedingConference contribution

6 Scopus citations

Abstract

Molecular Dynamics (MD) is an important atomistic simulation technique, with widespread use in computational chemistry, biology, and materials. An important limitation of MD is that the time step size is small, requiring a large number of iterations to simulate realistic time spans. Conventional parallelization is not very effective for this. We recently introduced a new approach to parallelization, where data from related prior simulations are used to parallelize a new computation along the time domain. In our prior work, the size of the physical system in the current simulation needed to be identical to that of the prior simulations. The significance of this paper lies in demonstrating a strategy that enables this approach to be used even when the physical systems differ in size. Furthermore, this method scaled up to almost 1000 processors with close to ideal speedup in one case, where conventional methods scale to only 2-3 processors.

Original languageEnglish (US)
Title of host publicationICPP 2006
Subtitle of host publicationProceedings of the 2006 International Conference on Parallel Processing
Pages119-126
Number of pages8
DOIs
StatePublished - 2006
Externally publishedYes
EventICPP 2006: 2006 International Conference on Parallel Processing - Columbus, OH, United States
Duration: Aug 14 2006Aug 18 2006

Publication series

NameProceedings of the International Conference on Parallel Processing
ISSN (Print)0190-3918

Other

OtherICPP 2006: 2006 International Conference on Parallel Processing
CountryUnited States
CityColumbus, OH
Period8/14/068/18/06

All Science Journal Classification (ASJC) codes

  • Hardware and Architecture
  • Engineering(all)

Fingerprint Dive into the research topics of 'Scalable time-parallelization of molecular dynamics simulations in nano mechanics'. Together they form a unique fingerprint.

Cite this