Method of moving frames to solve conservation laws on curved surfaces

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

A new numerical framework is proposed to solve partial differential equations on curved surfaces by using the orthogonal moving frames at each grid point to compute the gradient of a scalar variable. We call this framework the method of moving frames (MMF) that is adopted and modified from the works of É. Cartan. Compared to the Eulerian method and the Lagrangian multiplier method, the MMF method uses only the surface as the domain, not additionally the ambient space enclosing it. Also different from directly solving the equations with respect to the curved axis, the MMF method is free of the metric tensors. This uniqueness is the consequence of the virtual and penalty extension of the variables in a special direction, called the exponential direction, instead of the surface normal direction that is typically taken. The exponential extension eliminates the need to extend the computational domain and the variables outside the curved surfaces, but the variables outside the curved surfaces are not extended in the direction of the surface normal, yielding an extension error. However, the overall error for the MMF scheme, caused by the extension error, is of high order in L2 error with respect to space discretization. This high convergence rate implies that the exponential error can be made negligible compared to the error of differentiation and integration, which are also expressed with space discretization but with lower order, in adaptively-refined meshes proportional to the Gaussian curvature. As the first application of the MMF method, conservation laws are considered on curved surfaces. To display the exponential convergence and the unique features of the MMF scheme, convergence tests are demonstrated on four different types of surfaces: an open spherical shell, a closed spherical shell, an irregular closed surface, and a non-convex closed surface.

Original languageEnglish
Pages (from-to)268-294
Number of pages27
JournalJournal of Scientific Computing
Volume53
Issue number2
DOIs
Publication statusPublished - 2012 Nov

Fingerprint

Moving Frame
Curved Surface
Conservation Laws
Conservation
Normal Surface
Spherical Shell
Closed
Discretization
Lagrangian multiplier
Multiplier Method
Lagrangian Method
Exponential Convergence
Total curvature
Partial differential equations
Tensors
Penalty
Convergence Rate
Irregular
Eliminate
Uniqueness

All Science Journal Classification (ASJC) codes

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

Cite this

@article{1e5e01207be24619a769cdca5bbb4ff3,
title = "Method of moving frames to solve conservation laws on curved surfaces",
abstract = "A new numerical framework is proposed to solve partial differential equations on curved surfaces by using the orthogonal moving frames at each grid point to compute the gradient of a scalar variable. We call this framework the method of moving frames (MMF) that is adopted and modified from the works of {\'E}. Cartan. Compared to the Eulerian method and the Lagrangian multiplier method, the MMF method uses only the surface as the domain, not additionally the ambient space enclosing it. Also different from directly solving the equations with respect to the curved axis, the MMF method is free of the metric tensors. This uniqueness is the consequence of the virtual and penalty extension of the variables in a special direction, called the exponential direction, instead of the surface normal direction that is typically taken. The exponential extension eliminates the need to extend the computational domain and the variables outside the curved surfaces, but the variables outside the curved surfaces are not extended in the direction of the surface normal, yielding an extension error. However, the overall error for the MMF scheme, caused by the extension error, is of high order in L2 error with respect to space discretization. This high convergence rate implies that the exponential error can be made negligible compared to the error of differentiation and integration, which are also expressed with space discretization but with lower order, in adaptively-refined meshes proportional to the Gaussian curvature. As the first application of the MMF method, conservation laws are considered on curved surfaces. To display the exponential convergence and the unique features of the MMF scheme, convergence tests are demonstrated on four different types of surfaces: an open spherical shell, a closed spherical shell, an irregular closed surface, and a non-convex closed surface.",
author = "Sehun Chun",
year = "2012",
month = "11",
doi = "10.1007/s10915-011-9570-7",
language = "English",
volume = "53",
pages = "268--294",
journal = "Journal of Scientific Computing",
issn = "0885-7474",
publisher = "Springer New York",
number = "2",

}

Method of moving frames to solve conservation laws on curved surfaces. / Chun, Sehun.

In: Journal of Scientific Computing, Vol. 53, No. 2, 11.2012, p. 268-294.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Method of moving frames to solve conservation laws on curved surfaces

AU - Chun, Sehun

PY - 2012/11

Y1 - 2012/11

N2 - A new numerical framework is proposed to solve partial differential equations on curved surfaces by using the orthogonal moving frames at each grid point to compute the gradient of a scalar variable. We call this framework the method of moving frames (MMF) that is adopted and modified from the works of É. Cartan. Compared to the Eulerian method and the Lagrangian multiplier method, the MMF method uses only the surface as the domain, not additionally the ambient space enclosing it. Also different from directly solving the equations with respect to the curved axis, the MMF method is free of the metric tensors. This uniqueness is the consequence of the virtual and penalty extension of the variables in a special direction, called the exponential direction, instead of the surface normal direction that is typically taken. The exponential extension eliminates the need to extend the computational domain and the variables outside the curved surfaces, but the variables outside the curved surfaces are not extended in the direction of the surface normal, yielding an extension error. However, the overall error for the MMF scheme, caused by the extension error, is of high order in L2 error with respect to space discretization. This high convergence rate implies that the exponential error can be made negligible compared to the error of differentiation and integration, which are also expressed with space discretization but with lower order, in adaptively-refined meshes proportional to the Gaussian curvature. As the first application of the MMF method, conservation laws are considered on curved surfaces. To display the exponential convergence and the unique features of the MMF scheme, convergence tests are demonstrated on four different types of surfaces: an open spherical shell, a closed spherical shell, an irregular closed surface, and a non-convex closed surface.

AB - A new numerical framework is proposed to solve partial differential equations on curved surfaces by using the orthogonal moving frames at each grid point to compute the gradient of a scalar variable. We call this framework the method of moving frames (MMF) that is adopted and modified from the works of É. Cartan. Compared to the Eulerian method and the Lagrangian multiplier method, the MMF method uses only the surface as the domain, not additionally the ambient space enclosing it. Also different from directly solving the equations with respect to the curved axis, the MMF method is free of the metric tensors. This uniqueness is the consequence of the virtual and penalty extension of the variables in a special direction, called the exponential direction, instead of the surface normal direction that is typically taken. The exponential extension eliminates the need to extend the computational domain and the variables outside the curved surfaces, but the variables outside the curved surfaces are not extended in the direction of the surface normal, yielding an extension error. However, the overall error for the MMF scheme, caused by the extension error, is of high order in L2 error with respect to space discretization. This high convergence rate implies that the exponential error can be made negligible compared to the error of differentiation and integration, which are also expressed with space discretization but with lower order, in adaptively-refined meshes proportional to the Gaussian curvature. As the first application of the MMF method, conservation laws are considered on curved surfaces. To display the exponential convergence and the unique features of the MMF scheme, convergence tests are demonstrated on four different types of surfaces: an open spherical shell, a closed spherical shell, an irregular closed surface, and a non-convex closed surface.

UR - http://www.scopus.com/inward/record.url?scp=84869869362&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84869869362&partnerID=8YFLogxK

U2 - 10.1007/s10915-011-9570-7

DO - 10.1007/s10915-011-9570-7

M3 - Article

AN - SCOPUS:84869869362

VL - 53

SP - 268

EP - 294

JO - Journal of Scientific Computing

JF - Journal of Scientific Computing

SN - 0885-7474

IS - 2

ER -