Dispersion and dissipation errors of two fully discrete discontinuous Galerkin methods (Q1955922)

The authors analyze the dispersion and dissipation properties for Runge-Kutta discontinuous Galerkin (RKDG) methods and Lax-Wendroff discontinuous Galerkin (LWDG) methods for a linear advection equation. They focus on the fully discrete discontinuous Galerkin methods and derive analytical formulations of the dispersion and dissipation errors for both methods in terms of the Courant-Friedrichs-Lewy (CFL) number for \(kh\ll 1\) and carry out comparisons between these methods. In the first part of the paper, the authors introduce the above-mentioned two discontinuous Galerkin methods for a one-dimensional scalar conservation law. This is followed by an extensive analysis of the methods for the linear advection equation. Dissipation and dispersion errors are given in terms of \(K=kh\) where \(k\) is the wavenumber and \(h\) is the mesh size. Various spatial and temporal discretization schemes are considered for each method and their performances are compared. An alternative method for an analysis of the LWDG method (the fixed-\(\omega\) method) is carried out as well. Next, the authors extend their analysis for the case where the CFL number can depend on the number \(K\). The conclusion here is that with sufficiently small CFL numbers both methods lead to superconvergence in the dispersion and dissipation errors. In the final section of the paper, a set of numerical experiments is presented. Theoretical findings are verified by two different examples involving cosine wave and square-wave functions.