Fast parallel algorithms for short-range molecular dynamics (Q1346550): Difference between revisions
From MaRDI portal
Changed an Item |
Normalize DOI. |
||
(2 intermediate revisions by 2 users not shown) | |||
Property / DOI | |||
Property / DOI: 10.1006/jcph.1995.1039 / rank | |||
Property / MaRDI profile type | |||
Property / MaRDI profile type: MaRDI publication profile / rank | |||
Normal rank | |||
Property / OpenAlex ID | |||
Property / OpenAlex ID: W2019465613 / rank | |||
Normal rank | |||
Property / DOI | |||
Property / DOI: 10.1006/JCPH.1995.1039 / rank | |||
Normal rank |
Latest revision as of 18:32, 10 December 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Fast parallel algorithms for short-range molecular dynamics |
scientific article |
Statements
Fast parallel algorithms for short-range molecular dynamics (English)
0 references
5 April 1995
0 references
The paper presents three types of algorithms for the parallel computation of classical molecular dynamics. They are especially suited to short- range inter-molecular forces for which a cut-off range exists or can be imposed. One method allocates a defined subset of particles to each processor; another assigns each with a fixed subset of intermolecular forces and the last assigns a fixed spatial domain to each processor. The algorithms can be realized on distributed-memory machines with independent processors having suitable data flow capacities between them. Computations on a number of parallel computers are presented for a typical Lennard-Jones two-body potential, with sizes ranging from 500 up to 100,000,000 molecules. With current parallel machines the algorithms are seen to be competitive with the best vector Cray Y-MP and C90 algorithms, even in small size cases. For large problems the spatial algorithm can run at parallel efficiencies of 90\%. Relative merits of the algorithms and their application to more complicated dynamics are discussed.
0 references
short-range molecular dynamics
0 references
algorithms
0 references
parallel computation
0 references
Lennard-Jones two-body potential
0 references