An Adaptive Edge Element Method and Its Convergence for a Saddle‐Point Problem from Magnetostatics

Junqing Chen,Yifeng Xu,Jun Zou
DOI: https://doi.org/10.1002/num.20697
2011-01-01
Numerical Methods for Partial Differential Equations
Abstract:Reliable and efficient a posteriori error estimates are derived for the edge element discretization of a saddle‐point Maxwell's system. By means of the error estimates, an adaptive edge element method is proposed and its convergence is rigorously demonstrated. The algorithm uses a marking strategy based only on the error indicators, without the commonly used information on local oscillations and the refinement to meet the standard interior node property. Some new ingredients in the analysis include a novel quasi‐orthogonality and a new inf‐sup inequality associated with an appropriately chosen norm. It is shown that the algorithm is a contraction for the sum of the energy error plus the error indicators after each refinement step. Numerical experiments are presented to show the robustness and effectiveness of the proposed adaptive algorithm. © 2011 Wiley Periodicals, Inc. Numer Methods Partial Differential Eq, 2012
What problem does this paper attempt to address?