A new Lax-Wendroff discontinuous Galerkin method with superconvergence (Q898429): Difference between revisions

The authors consider hyperbolic conservation laws, especially the 1D and 2D linear advection equation and the 1D Euler system. First they explain their way of discretization (for the Lax-Wendroff approach to reexpress the time derivatives in a Taylor expansion in time through spatial derivatives, following Qiu, Dumbser and Shu [\textit{J. Qiu} et al., Comput. Methods Appl. Mech. Eng. 194, No. 42--44, 4528--4543 (2005; Zbl 1093.76038)]; this is also known as Cauchy-Kobalewski-type time discretization by \textit{Ju. I. Sokin} [Tr. Mat. Inst. Steklova 122, 66--84 (1973; Zbl 0292.35052)] and combining this with the local discontinuous Galerkin (LDG) method (passing to a system of equations and discretizing it in space, taking care of the approximation of fluxes appearing in this process). Then they show that the semi-discrete LDG method for 1D nonlinear scalar conservation laws (with appropriate numerical fluxes) is \(L_2\)-stable and hint on the possibility to enhance the order of the numerical solution up to \(2k+1\) when taking \(P^k\) polynomials in space and using the postprocessing of Cockburn, Luskin, Shu an Süli [\textit{B. Cockburn} et al., Math. Comput. 72, No. 242, 577--606 (2003; Zbl 1015.65049)]. This is then illustrated by numerical experiments to hold true for their method which does not need higher-order stages like the concurrent SSP RKDG schemes (but the present authors need additional discrete unknowns due to the passing to systems of equations). The superconvergence is also backed by a Fourier analysis based on symbolic computations for \(k=1, 2\), the 1D linear advection equation, and for small Courant numbers.