Diffusive relaxation schemes for multiscale discrete-velocity kinetic equations

Shi Jin, Lorenzo Pareschi, Giuseppe Toscani

Research output: Contribution to journalArticlepeer-review

140 Scopus citations

Abstract

Many kinetic models of the Boltzmann equation have a diffusive scaling that leads to the Navier-Stokes-type parabolic equations, such as the heat equation, the porous media equations, the advection-diffusion equation, and the viscous Burgers equation. In such problems the diffusive relaxation parameter may differ in several orders of magnitude from the rarefied regimes to the hydrodynamic (diffusive) regimes, and it is desirable to develop a class of numerical schemes that can work uniformly with respect to this relaxation parameter. Earlier approaches that work from the rarefied regimes to the Euler regimes do not directly apply to these problems since here, in addition to the stiff relaxation term, the convection term is also stiff. Our idea is to reformulate the problem in the form commonly used for the relaxation schemes to conservation laws by properly combining the stiff component of the convection terms into the relaxation term. This, however, introduces new difficulties due to the dependence of the stiff source term on the gradient. We show how to overcome this new difficulty with an adequately designed, economical discretization procedure for the relaxation term. These schemes are shown to have the correct diffusion limit. Several numerical results in one and two dimensions are presented, which show the robustness, as well as the uniform accuracy, of our schemes.

Original languageEnglish
Pages (from-to)2405-2439
Number of pages35
JournalSIAM Journal on Numerical Analysis
Volume35
Issue number6
DOIs
StatePublished - 1998
Externally publishedYes

Keywords

  • Diffusion limit
  • Discrete-velocity kinetic models
  • Relaxation schemes
  • Stiff terms

Fingerprint

Dive into the research topics of 'Diffusive relaxation schemes for multiscale discrete-velocity kinetic equations'. Together they form a unique fingerprint.

Cite this