A methodology towards automatic implementation of N-body algorithms
Applied Numerical Mathematics - Applied and computational mathematics: Selected papers of the third panamerican workshop Trujillo, Peru, 24-28 April 2000
A Bayesian Approach to Joint Feature Selection and Classifier Design
IEEE Transactions on Pattern Analysis and Machine Intelligence
A fast method for solving the heat equation by layer potentials
Journal of Computational Physics
Proceedings of the 2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis
Hi-index | 0.01 |
We present a matrix representation for the fast Gauss transform (FGT) originally proposed by Greengard and Strain. With the matrix representation we reveal the matrix structures explored and exploited in the FGT, relate the multidimensional FGT to the one-dimensional FGT via Kronecker products, and unify various FGT versions. Based on the unifying representation, we present also a framework of FGT algorithms that demonstrates an algorithmic approach to utilizing the revealed matrix factor structures and suggests computational varieties for adapting the FGT to architecture specifics as well as application specifics to achieve optimal performance.