Multilevel Algorithms for Large-scope Molecular Dynamics Simulations of Nanostructures on Parallel Computers
Author(s) -
Aiichiro Nakano,
Rajiv K. Kalia,
Priya Vashishta
Publication year - 1998
Publication title -
vlsi design
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.123
H-Index - 24
eISSN - 1065-514X
pISSN - 1026-7123
DOI - 10.1155/1998/93670
Subject(s) - curvilinear coordinates , computer science , abstraction , conjugate gradient method , molecular dynamics , algorithm , fuzzy logic , theoretical computer science , computational science , mathematics , chemistry , artificial intelligence , computational chemistry , geometry , philosophy , epistemology
Molecular Dynamics (MD) is a powerful tool for the atomistic understanding of longrangestress-mediated phenomena, phonon properties, and mechanical failure ofnanostructures. For realistic modeling of nanostructures, however, the scope ofsimulations must be extended to large system sizes, long simulated times, and complexrealism. We have developed new multilevel algorithms and physical models encompassingmultiple levels of abstraction: i) space-time multiresolution schemes; ii) adaptivecurvilinear-coordinate load balancing; iii) hierarchical dynamics via a rigid-body/implicit-integration/normal-mode approach; iv) variable-charge MD based on electronegativityequalization; and v) multilevel preconditioned conjugate gradient method.Fuzzy clustering is used to facilitate the seamless integration of the multiple levels ofabstraction
Accelerating Research
Robert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom
Address
John Eccles HouseRobert Robinson Avenue,
Oxford Science Park, Oxford
OX4 4GP, United Kingdom