A Rigid-Body-Based Multiple Time Scale Molecular Dynamics Simulation of Nanophase Materials

Authors:
Aiichiro Nakano
Affiliations:
Department of Computer Science, Concurrent Computing Laboratory for Materials Simulations, Louisiana State University, Baton Rouge, Louisiana, U.S.A.
Venue:
International Journal of High Performance Computing Applications
Year:
1999

Citing 6
Cited 3

Computer simulation of liquids

Computer simulation of liquids
A fast recursive algorithm for molecular dynamics simulation

Journal of Computational Physics
LIN: a new algorithm to simulate the dynamics of biomolecules by combining implicit-integration and normal mode techniques

Journal of Computational Chemistry
Algorithms and software for ordinary differential equations and differential-algebraic equations, part II: higher-order methods and software packages

Computers in Physics
A Family of Symplectic Integrators: Stability, Accuracy, and Molecular Dynamics Applications

SIAM Journal on Scientific Computing
An adaptive curvilinear-coordinate approach to dynamic load balancing of parallel multiresolution molecular dynamics

Parallel Computing

Scalable atomistic simulation algorithms for materials research

Proceedings of the 2001 ACM/IEEE conference on Supercomputing
Scalable Molecular-Dynamics, Visualization, and Data-Management Algorithms for Materials Simulations

Computing in Science and Engineering
Scalable atomistic simulation algorithms for materials research

Scientific Programming - Best papers from SC 2001

Quantified Score

Hi-index	0.00

Visualization

Abstract

Nanophase technology achieves superior material properties by assembling nanometer-size clusters. Structures on multiple-length scales and a hierarchy of time scales are essential for the design and control of nanophase materials. However, coexistence of a wide range of length and time scales hinders atomistic simulations of these materials. A new algorithm is developed for large-scale, longtime molecular dynamics simulations by combining (1) quaternion-based, rigid-body dynamics for global cluster motions; (2) implicit integration of Newton's equations for the coalescence of clusters; and (3) normal-mode analysis of fast atomic oscillations. The new scheme, using a time step &Dgr;t of 10-12 seconds, speeds up a conventional explicit integration scheme (&Dgr;t=2 × 10-15 seconds) by a factor of 28. A parallel implementation of the scheme achieves an efficiency of 0.94 for a 1.28-million-atom nanocrystalline silicon nitride solid on 64 nodes of an IBM SP computer.