Residual estimates for post-processors in elliptic problems (Q2049081)

A strategy is developed for improving existing post-processing strategies for numerical solutions of a model elliptic problem by controlling the a posteriori error. This strategy is based on a modification of the post-processed the solution so that it satisfies Galerkin orthogonality. Various a priori type error estimates show the convergence properties of the orthogonal post-processor including an increased order of accuracy in the \(L_2\) norm. The convergence analysis is supported with numerical examples using the smoothness increasing accuracy conserving (SIAC) filter and superconvergent patch recovery (SPR) processors approximating smooth and non-smooth solutions. Furthermore, a reliable and efficient a posteriori error estimator is constructed for the orthogonal post-processed solution. Extensive numerical tests are performed on the SIAC and SPR post-processors to show the performance of the a posteriori bounds, the effect of smoothness of the solution on the post-processors and to study adaptive methods driven by these estimators.