High order filtering methods for approximating hyperbolic systems of conservation laws (Q1175640): Difference between revisions

Recently, a new class of high order schemes related to numerical calculations of linear and nonlinear systems of conservation laws has been developed. This new class of methods uses an explicit total variation diminishing Runge-Kutta multistage time discretization together with high order essentially non-oscillatory (ENO) spatial discretization methods. However, the recently developed ENO schemes are computationally costly compared to simple central difference methods. In this paper the authors develop a filtering method which uses simple central differencing of arbitrarily high order accuracy, except when a novel local test indicates the development of spurious oscillations. At these points, generally few in number, the authors use the full ENO apparatus, maintaining (in fact, improving) the high order of accuracy, but removing spurious oscillations. Section 2 is devoted to a detailed presentation of the method. Section 3 offers several numerical examples in one and two space dimensions for both scalars and systems of conservation laws, indicating the success of the method. In addition, a vectorized code of the filtering method is implemented and proven to be computationally very efficient. Section 4 concludes the paper and proposes several different approaches for solving nonlinear problems in general domains using finite element grids.