Error estimates and post-processing of local discontinuous Galerkin method for Schrödinger equations (Q2423678)

The authors consider the approximation of the Schrödinger equation with variable coefficients using the local discontinuous Galerkin (LDG) method. Error estimates are proved in \(L^2\) and negative-order norms. The order is optimal in the \(L^2\)-norm whereas it is high in the negative order norm. Using the smoothness-increasing accuracy-conserving (SIAC) filter, the accuracy of the LDG approximation can be improved from \(\mathcal{O} (h^{k + 1})\) (when polynomials of degree less or equal to \(k\) are used) to at least \(\mathcal{O}(h^{2k})\), when the alternating numerical fluxes are adopted in the method. Some numerical tests are presented. It is shown numerically that the accuracy of LDG solutions can be improved from \(\mathcal{O}(h^{k + 1})\) to at least \(\mathcal{O}(h^{2 k + 1})\) using alternating numerical fluxes.