Properties-preserving high order numerical methods for a kinetic eikonal equation

详细信息    查看全文

• 作者：Songting Luo ; ^{luos@iastate.edu} ; Nicholas Payne ^{napayne@iastate.edu}
• 关键词：BGK equation ; Hopf&ndash ; Cole transformation ; Kinetic eikonal equation ; WENO ; Runge&ndash ; Kutta ; Godunov ; Properties-preserving
• 刊名：Journal of Computational Physics
• 出版年：2017
• 出版时间：15 February 2017
• 年：2017
• 卷：331
• 期：Complete
• 页码：73-89
• 全文大小：2149 K
• 卷排序：331

文摘

For the BGK (Bhatnagar–Gross–Krook) equation in the large scale hyperbolic limit, the density of particles can be transformed as the Hopf–Cole transformation, where the phase function converges uniformly to the viscosity solution of an effective Hamilton–Jacobi equation, referred to as the kinetic eikonal equation. In this work, we present efficient high order finite difference methods for numerically solving the kinetic eikonal equation. The methods are based on monotone schemes such as the Godunov scheme. High order weighted essentially non-oscillatory techniques and Runge–Kutta procedures are used to obtain high order accuracy in both space and time. The effective Hamiltonian is determined implicitly by a nonlinear equation given as integrals with respect to the velocity variable. Newton's method is applied to solve the nonlinear equation, where integrals with respect to the velocity variable are evaluated either by a Gauss quadrature formula or as expansions with respect to moments of the Maxwellian. The methods are designed such that several key properties such as the positivity of the viscosity solution and the positivity of the effective Hamiltonian are preserved. Numerical experiments are presented to demonstrate the effectiveness of the methods.