Asymptotically exact functional error estimators based on superconvergent gradient recovery (Q818586)

The use of dual/adjoint problems for approximating functionals of solutions of partial differential equations with good accuracy or to derive a goal-oriented adaptive refinement scheme is an active area of research. The author presents two new functional error estimators and gives conditions in order to be asymptotically exact. The first is of dual residual weighting type and is derived for meshes in which most triangles satisfy an \(O(h^2)\)-approximate parallelogram property. The second functional estimator involves dual error estimate weighting using any superconvergent gradient recovery technique for the primal and dual solutions. Numerical experiments are carried out demonstrating the asymptotic exactness of a dual error estimate weighting estimator and the effectiveness of refinement done with respect to the local error indicators.