The fast generalized Gauss transform

Marina Spivak, Shravan K. Veerapaneni, Leslie Greengard

Research output: Contribution to journalArticle

Abstract

The fast Gauss transform allows for the calculation of the sum of N Gaussians at M points in O(N + M) time. Here, we extend the algorithm to a wider class of kernels, motivated by quadrature issues that arise in using integral equation methods for solving the heat equation on moving domains. In particular, robust high-order product integration methods require convolution with O(q) distinct Gaussian-type kernels in order to obtain qth-order accuracy in time. The generalized Gauss transform permits the computation of each of these kernels and, thus, the construction of fast solvers with optimal computational complexity. We also develop plane-wave representations of these Gaussian-type fields, permitting the "diagonal translation" version of the Gauss transform to be applied. When the sources and targets lie on a uniform grid, or a hierarchy of uniform grids, we show that the curse of dimensionality (the exponential growth of complexity constants with dimension) can be mitigated. Under these conditions, the algorithm has a lower operation count than the fast Fourier transform even for modest values of N and M.

Original languageEnglish (US)
Pages (from-to)3092-3107
Number of pages16
JournalSIAM Journal on Scientific Computing
Volume32
Issue number5
DOIs
StatePublished - 2010

Fingerprint

Gauss Transform
kernel
Fast Solvers
Convolution
Fast Fourier transforms
Product Integration
Integral equations
Grid
Computational complexity
Integral Equation Method
Curse of Dimensionality
Exponential Growth
Fast Fourier transform
Quadrature
Heat Equation
Plane Wave
Count
Computational Complexity
Higher Order
Distinct

Keywords

  • Fast algorithms
  • Gauss transform
  • Heat potentials
  • High-order accuracy
  • Tensorproduct grids

ASJC Scopus subject areas

  • Applied Mathematics
  • Computational Mathematics

Cite this

The fast generalized Gauss transform. / Spivak, Marina; Veerapaneni, Shravan K.; Greengard, Leslie.

In: SIAM Journal on Scientific Computing, Vol. 32, No. 5, 2010, p. 3092-3107.

Research output: Contribution to journalArticle

Spivak, Marina ; Veerapaneni, Shravan K. ; Greengard, Leslie. / The fast generalized Gauss transform. In: SIAM Journal on Scientific Computing. 2010 ; Vol. 32, No. 5. pp. 3092-3107.
@article{eb42e6272fc3480b9e88f297da439889,
title = "The fast generalized Gauss transform",
abstract = "The fast Gauss transform allows for the calculation of the sum of N Gaussians at M points in O(N + M) time. Here, we extend the algorithm to a wider class of kernels, motivated by quadrature issues that arise in using integral equation methods for solving the heat equation on moving domains. In particular, robust high-order product integration methods require convolution with O(q) distinct Gaussian-type kernels in order to obtain qth-order accuracy in time. The generalized Gauss transform permits the computation of each of these kernels and, thus, the construction of fast solvers with optimal computational complexity. We also develop plane-wave representations of these Gaussian-type fields, permitting the {"}diagonal translation{"} version of the Gauss transform to be applied. When the sources and targets lie on a uniform grid, or a hierarchy of uniform grids, we show that the curse of dimensionality (the exponential growth of complexity constants with dimension) can be mitigated. Under these conditions, the algorithm has a lower operation count than the fast Fourier transform even for modest values of N and M.",
keywords = "Fast algorithms, Gauss transform, Heat potentials, High-order accuracy, Tensorproduct grids",
author = "Marina Spivak and Veerapaneni, {Shravan K.} and Leslie Greengard",
year = "2010",
doi = "10.1137/100790744",
language = "English (US)",
volume = "32",
pages = "3092--3107",
journal = "SIAM Journal of Scientific Computing",
issn = "1064-8275",
publisher = "Society for Industrial and Applied Mathematics Publications",
number = "5",

}

TY - JOUR

T1 - The fast generalized Gauss transform

AU - Spivak, Marina

AU - Veerapaneni, Shravan K.

AU - Greengard, Leslie

PY - 2010

Y1 - 2010

N2 - The fast Gauss transform allows for the calculation of the sum of N Gaussians at M points in O(N + M) time. Here, we extend the algorithm to a wider class of kernels, motivated by quadrature issues that arise in using integral equation methods for solving the heat equation on moving domains. In particular, robust high-order product integration methods require convolution with O(q) distinct Gaussian-type kernels in order to obtain qth-order accuracy in time. The generalized Gauss transform permits the computation of each of these kernels and, thus, the construction of fast solvers with optimal computational complexity. We also develop plane-wave representations of these Gaussian-type fields, permitting the "diagonal translation" version of the Gauss transform to be applied. When the sources and targets lie on a uniform grid, or a hierarchy of uniform grids, we show that the curse of dimensionality (the exponential growth of complexity constants with dimension) can be mitigated. Under these conditions, the algorithm has a lower operation count than the fast Fourier transform even for modest values of N and M.

AB - The fast Gauss transform allows for the calculation of the sum of N Gaussians at M points in O(N + M) time. Here, we extend the algorithm to a wider class of kernels, motivated by quadrature issues that arise in using integral equation methods for solving the heat equation on moving domains. In particular, robust high-order product integration methods require convolution with O(q) distinct Gaussian-type kernels in order to obtain qth-order accuracy in time. The generalized Gauss transform permits the computation of each of these kernels and, thus, the construction of fast solvers with optimal computational complexity. We also develop plane-wave representations of these Gaussian-type fields, permitting the "diagonal translation" version of the Gauss transform to be applied. When the sources and targets lie on a uniform grid, or a hierarchy of uniform grids, we show that the curse of dimensionality (the exponential growth of complexity constants with dimension) can be mitigated. Under these conditions, the algorithm has a lower operation count than the fast Fourier transform even for modest values of N and M.

KW - Fast algorithms

KW - Gauss transform

KW - Heat potentials

KW - High-order accuracy

KW - Tensorproduct grids

UR - http://www.scopus.com/inward/record.url?scp=78149337561&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78149337561&partnerID=8YFLogxK

U2 - 10.1137/100790744

DO - 10.1137/100790744

M3 - Article

AN - SCOPUS:78149337561

VL - 32

SP - 3092

EP - 3107

JO - SIAM Journal of Scientific Computing

JF - SIAM Journal of Scientific Computing

SN - 1064-8275

IS - 5

ER -