Stability of Radial Basis Function Methods for Convection Problems on the Circle and Sphere

Jordan M. Martel, Rodrigo Platte

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

This paper investigates the stability of the radial basis function (RBF) collocation method for convective problems on the circle and sphere. We prove that the RBF method is Lax-stable for problems on the circle when the collocation points are equispaced and the transport speed is constant. We also show that the eigenvalues of discretization matrices are purely imaginary in the case of variable coefficients and equispaced nodes. By studying the (Formula presented.)-pseudospectra of these matrices we argue that approximations are also Lax-stable in the latter case. Based on these results, we conjecture that the discretization of transport operators on the sphere present a similar behavior. We provide strong evidence that the method is Lax-stable on the sphere when the collocation points come from certain polyhedra. In both geometries, we demonstrate that eigenvalues of the differentiation matrix deviate from the imaginary axis linearly with perturbations off the set of ideal collocation points. When the ideal set is impractical or unavailable, we propose a least-squares method and present numerical evidence suggesting that it can substantially improve stability without any increase to computational cost and with only a minor cost to accuracy.

Original languageEnglish (US)
Pages (from-to)1-19
Number of pages19
JournalJournal of Scientific Computing
DOIs
StateAccepted/In press - Apr 15 2016

Fingerprint

Collocation
Radial Functions
Convection
Basis Functions
Circle
Discretization
Pseudospectra
Eigenvalue
Collocation Method
Variable Coefficients
Least Square Method
Polyhedron
Computational Cost
Costs
Minor
Linearly
Perturbation
Geometry
Approximation
Vertex of a graph

Keywords

  • Eigenvalue stability
  • Kernel methods
  • Periodic functions
  • Pseudospectra
  • RBF
  • Spectral methods

ASJC Scopus subject areas

  • Software
  • Computational Theory and Mathematics
  • Theoretical Computer Science
  • Engineering(all)

Cite this

Stability of Radial Basis Function Methods for Convection Problems on the Circle and Sphere. / Martel, Jordan M.; Platte, Rodrigo.

In: Journal of Scientific Computing, 15.04.2016, p. 1-19.

Research output: Contribution to journalArticle

@article{c95827fa4b2c45f2b34d8bb4e490d572,
title = "Stability of Radial Basis Function Methods for Convection Problems on the Circle and Sphere",
abstract = "This paper investigates the stability of the radial basis function (RBF) collocation method for convective problems on the circle and sphere. We prove that the RBF method is Lax-stable for problems on the circle when the collocation points are equispaced and the transport speed is constant. We also show that the eigenvalues of discretization matrices are purely imaginary in the case of variable coefficients and equispaced nodes. By studying the (Formula presented.)-pseudospectra of these matrices we argue that approximations are also Lax-stable in the latter case. Based on these results, we conjecture that the discretization of transport operators on the sphere present a similar behavior. We provide strong evidence that the method is Lax-stable on the sphere when the collocation points come from certain polyhedra. In both geometries, we demonstrate that eigenvalues of the differentiation matrix deviate from the imaginary axis linearly with perturbations off the set of ideal collocation points. When the ideal set is impractical or unavailable, we propose a least-squares method and present numerical evidence suggesting that it can substantially improve stability without any increase to computational cost and with only a minor cost to accuracy.",
keywords = "Eigenvalue stability, Kernel methods, Periodic functions, Pseudospectra, RBF, Spectral methods",
author = "Martel, {Jordan M.} and Rodrigo Platte",
year = "2016",
month = "4",
day = "15",
doi = "10.1007/s10915-016-0206-9",
language = "English (US)",
pages = "1--19",
journal = "Journal of Scientific Computing",
issn = "0885-7474",
publisher = "Springer New York",

}

TY - JOUR

T1 - Stability of Radial Basis Function Methods for Convection Problems on the Circle and Sphere

AU - Martel, Jordan M.

AU - Platte, Rodrigo

PY - 2016/4/15

Y1 - 2016/4/15

N2 - This paper investigates the stability of the radial basis function (RBF) collocation method for convective problems on the circle and sphere. We prove that the RBF method is Lax-stable for problems on the circle when the collocation points are equispaced and the transport speed is constant. We also show that the eigenvalues of discretization matrices are purely imaginary in the case of variable coefficients and equispaced nodes. By studying the (Formula presented.)-pseudospectra of these matrices we argue that approximations are also Lax-stable in the latter case. Based on these results, we conjecture that the discretization of transport operators on the sphere present a similar behavior. We provide strong evidence that the method is Lax-stable on the sphere when the collocation points come from certain polyhedra. In both geometries, we demonstrate that eigenvalues of the differentiation matrix deviate from the imaginary axis linearly with perturbations off the set of ideal collocation points. When the ideal set is impractical or unavailable, we propose a least-squares method and present numerical evidence suggesting that it can substantially improve stability without any increase to computational cost and with only a minor cost to accuracy.

AB - This paper investigates the stability of the radial basis function (RBF) collocation method for convective problems on the circle and sphere. We prove that the RBF method is Lax-stable for problems on the circle when the collocation points are equispaced and the transport speed is constant. We also show that the eigenvalues of discretization matrices are purely imaginary in the case of variable coefficients and equispaced nodes. By studying the (Formula presented.)-pseudospectra of these matrices we argue that approximations are also Lax-stable in the latter case. Based on these results, we conjecture that the discretization of transport operators on the sphere present a similar behavior. We provide strong evidence that the method is Lax-stable on the sphere when the collocation points come from certain polyhedra. In both geometries, we demonstrate that eigenvalues of the differentiation matrix deviate from the imaginary axis linearly with perturbations off the set of ideal collocation points. When the ideal set is impractical or unavailable, we propose a least-squares method and present numerical evidence suggesting that it can substantially improve stability without any increase to computational cost and with only a minor cost to accuracy.

KW - Eigenvalue stability

KW - Kernel methods

KW - Periodic functions

KW - Pseudospectra

KW - RBF

KW - Spectral methods

UR - http://www.scopus.com/inward/record.url?scp=84963777436&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84963777436&partnerID=8YFLogxK

U2 - 10.1007/s10915-016-0206-9

DO - 10.1007/s10915-016-0206-9

M3 - Article

AN - SCOPUS:84963777436

SP - 1

EP - 19

JO - Journal of Scientific Computing

JF - Journal of Scientific Computing

SN - 0885-7474

ER -