Two-grid discontinuous Galerkin method for quasi-linear elliptic problems (Q429959)

The authors consider a quasilinear two-dimensional elliptic equation (containing a nonlinear diffusion coefficient) along with Dirichlet boundary conditions. They propose the SIPG version of the discontinuous Galerkin method for the numerical solution in combination with two grids characterized by the mesh parameters \(h,H\), where \(h\ll H\ll 1\). On the coarse grid a fixed point iteration is envisaged but on the fine grid (which is embedded into the coarse grid) the argument of the nonlinear diffusion coefficient is the coarse-grid solution. In this way, a linear problem remains on the fine grid. In their accuracy estimates they base on work of \textit{T. Gudi} and \textit{A. K. Pani} [SIAM J. Numer. Anal. 45, No. 1, 163--192 (2007; Zbl 1140.65082)], and of \textit{D. N. Arnold} et al. [SIAM J. Numer. Anal. 39, No. 5, 1749--1779 (2002; Zbl 1008.65080)], between others, and prove an accuracy of order \(h^r+H^{r+1}\) in the broken \(H^1\) norm where \(r\) is the polynomial degree of the finite element space, assuming a nonlinearity which is bounded in \(C^1\). The numerical experiments (using a simple iteration on the coarse grid) illustrate and verify the estimates and the theoretical outcome that the solution of the nonlinear problem is not much more involving than that of a linear one, from the point of view of the fine grid -- a conclusion well known from nonlinear multigrid. In an appendix, they investigate in detail pointwise error estimates for the three-dimensional case so that the results hold also in \(\mathbb R^3\).