Superconvergent postprocessing of the continuous Galerkin time stepping method for nonlinear initial value problems with application to parabolic problems (Q6101869)

The authors extend the superconvergent postprocessing techniques introduced in [\textit{M. Baccouch}, Appl. Numer. Math. 106, 129--153 (2016; Zbl 1382.65220)] for the Discontinuous Galerkin (DG) time stepping method to the Continuous Galerkin (CG) case. For this end and for clarity, a single nonlinear first-order initial value problem (IVP) is first considered and the postprocessing technique proposed in this paper is well described. The results can be extended to systems of such ODEs. The main idea of the postprocessing technique is to add a higher order Lobatto polynomial of degree \(k+1\) to the CG solution of degree \(k\). Global superconvergent error bounds for the postprocessed CG approximations over arbitrary time partitions are first established, and as a by product, it is shown that the convergence rates of the \(L^2\)-, \(H^1\)- and \(L^\infty\)-estimates for the CG method with quasi-uniform meshes for smooth solutions are improved by one order. For solutions with initial singularities, it is proved that the optimal global error estimates and nodal superconvergent estimate can be obtained for the CG method with graded meshes, and after postprocessing, the convergence rates of the \(L^2\)-, \(H^1\)- and \(L^\infty\)-estimates are also improved by one order. The superconvergent postprocessing technique is applied to the CG time discretization of nonlinear parabolic equations. Some numerical examples are presented to support the theoretical results.