Adaptive discontinuous Galerkin methods on surfaces (Q2634614)

The authors derive a dual weighted residual-based a posteriori error estimate for a surface discontinuous Galerkin (DG) discretization of a model second-order elliptic problem posed on a smooth surface in \(\mathbb{R}^3\). Both reliability and efficiency of the error estimator in the energy norm are proved. It is shown that the error may be split into a residual part, made up of the standard residual term along with the jump of the DG approximation, and a higher-order geometric part which arises from the lack of Galerkin orthogonality. Upper and lower bounds for the resulting a posteriori error estimator are proved. Some challenging test problems to demonstrate the reliability and efficiency of the estimator are considered. A novel geometric driven refinement strategy for partial differential equations on surfaces which considerably improves the performance of the method on complex surfaces is also presented.