Compact and stable discontinuous Galerkin methods for convection-diffusion problems (Q2882789)

The authors consider the following convection-diffusion problem: NEWLINE\[NEWLINE\begin{cases}\partial_t u+\nabla \cdot (f(u)-A(u)\nabla u)=s(u) & \text{in } \Omega \times (0,t_{end}), \\ u=g_D &\text{on } \partial \Omega \times [0,t_{end}), \\ u(0,\cdot)=u_0 & \text{in } \Omega,\end{cases}NEWLINE\]NEWLINE where \(\Omega \subset \mathbb{R}^d\) with \(d \in \{2,3\}\) is bounded polygonal, \(A:\mathbb{R} \to \mathbb{R}^{d \times d}, ~f:\mathbb{R} \to \mathbb{R}^d\) and \(s:\mathbb{R} \to \mathbb{R}\). The diffusion term is discretized by the new compact discontinuous Galerkin 2 scheme (CDG2). One main feature of the CGD2 method is the compactness of the stencil which includes only neighbouring elements, even for higher order approximation.NEWLINENEWLINEThe authors prove coercivity and stability of their scheme for the Poisson and the heat equation and they provide numerical tests for a linear convection-diffusion and the compressible nonlinear Navier-Stokes equation comparing their method with other existing discontinuous Galerkin methods.