A methodology towards automatic implementation of N-body algorithms

Authors:
V. P. Pauca;A. F. Rodriguez;X. Sun;K. S. Trivedi
Affiliations:
Department of Computer Science, Box 90129, Duke University, Durham, NC;Department of Computer Science, Box 90129, Duke University, Durham, NC and IBM (ITSO), 1001 Winstead Drive, Cary, NC;Department of Computer Science, Box 90129, Duke University, Durham, NC;Department of Electrical and Computer Engineering, Box 90291, Duke University, Durham, NC
Venue:
Applied Numerical Mathematics - Applied and computational mathematics: Selected papers of the third panamerican workshop Trujillo, Peru, 24-28 April 2000
Year:
2002

Citing 12
Cited 1

A fast algorithm for particle simulations

Journal of Computational Physics
The numerical solution of the N-body problem

Computers in Physics
The fast Gauss transform

SIAM Journal on Scientific and Statistical Computing
An implementation of the fast multipole method without multipoles

SIAM Journal on Scientific and Statistical Computing
The parallel fast multipole method in molecular dynamics

The parallel fast multipole method in molecular dynamics
Matrix computations (3rd ed.)

Matrix computations (3rd ed.)
A method of images for the evaluation of electrostatic fields in systems of closely spaced conducting cylinders

SIAM Journal on Applied Mathematics
Modification of the carrier, Greengard, and Rokhlin FMM for independent source and target fields

Journal of Computational Physics
A Matrix Version of the Fast Multipole Method

SIAM Review
A Kronecker Product Representation of the Fast Gauss Transform

SIAM Journal on Matrix Analysis and Applications
An {\it bf O(N)} Algorithm for Three-Dimensional N-body Simulations

An {\'it bf O(N)} Algorithm for Three-Dimensional N-body Simulations
Efficient parallel implementations of multipole based n-body algorithms

Efficient parallel implementations of multipole based n-body algorithms

Automating the development of quantum computational software

ACM-SE 45 Proceedings of the 45th annual southeast regional conference

Quantified Score

Hi-index	0.00

Visualization

Abstract

We propose a methodology aimed at automating the software development of fast discrete transforms for N-body problems. The methodology starts with a representation of the transform matrix in compact form. Then, two translation phases are applied. One translates matrix notation to an intermediate form. The other translates the intermediate form to compilable code for a target computing environment. The generated target code employs a library of pre-defined and optimized routines. We apply our methodology systematically to implement the three-dimensional fast multipole method. We demonstrate the efficiency of our code in a molecular dynamics simulation.