Second-order methods for diffusion-convection equations
✍ Scribed by Kohler, Theresa ;Voss, David
- Publisher
- John Wiley and Sons
- Year
- 1999
- Tongue
- English
- Weight
- 141 KB
- Volume
- 15
- Category
- Article
- ISSN
- 1069-8299
No coin nor oath required. For personal study only.
✦ Synopsis
The method of lines provides a ¯exible and general approach for solving time-dependent PDEs. However, the numerical solution of the resulting ODE system can present certain diculties depending on the method used. In particular, oscillations may appear in the solution when standard methods are applied to the ODE system arising from the semi-discretization of the diusion±convection equation duadt ad 2 uadx 2 Àbduadx. We examine second-order methods for such systems and present economical L-stable predictor± corrector schemes which are oscillation-free.
📜 SIMILAR VOLUMES
This paper presents a characteristic Galerkin "nite element method with an implicit algorithm for solving multidimensional, time-dependent convection}di!usion equations. The method is formulated on the basis of the combination of both the precise and the implicit numerical integration procedures aim
Some modi®ed AGE methods for the convection±diusion equation are developed in this paper. Firstly, there is a treatment on the convection term in the equation which is dierent from that in the AGE method by Evans and Abdullah (1985). Secondly, upwind-type schemes are used for the convection dominate
A proof of high-order convergence of three deterministic particle methods for the convectiondiffusion equation in two dimensions is presented. The methods are based on discretizations of an integro-differential equation in which an integral operator approximates the diffusion operator. The methods d
In this paper we analyze convergence of basic iterative Jacobi and Gauss-Seidel type methods for solving linear systems which result from finite element or finite volume discretization of convection-diffusion equations on unstructured meshes. In general the resulting stiffness matrices are neither M
The solution of the linear system Ax = b by iterative methods requires a splitting of the coefficient matrix in the form A = M -N where M is usually chosen to be a diagonal or a triangular matrix. In this article we study relaxation methods induced by the Hermitian and skew-Hermitian splittings for