Adjoint Based A Posteriori Analysis of Multiscale Mortar Discretizations with Multinumerics
KAUST Grant NumberKUK-C1-013-04
MetadataShow full item record
AbstractIn this paper we derive a posteriori error estimates for linear functionals of the solution to an elliptic problem discretized using a multiscale nonoverlapping domain decomposition method. The error estimates are based on the solution of an appropriately defined adjoint problem. We present a general framework that allows us to consider both primal and mixed formulations of the forward and adjoint problems within each subdomain. The primal subdomains are discretized using either an interior penalty discontinuous Galerkin method or a continuous Galerkin method with weakly imposed Dirichlet conditions. The mixed subdomains are discretized using Raviart- Thomas mixed finite elements. The a posteriori error estimate also accounts for the errors due to adjoint-inconsistent subdomain discretizations. The coupling between the subdomain discretizations is achieved via a mortar space. We show that the numerical discretization error can be broken down into subdomain and mortar components which may be used to drive adaptive refinement.Copyright © by SIAM.
CitationTavener S, Wildey T (2013) Adjoint Based A Posteriori Analysis of Multiscale Mortar Discretizations with Multinumerics. SIAM Journal on Scientific Computing 35: A2621–A2642. Available: http://dx.doi.org/10.1137/12089973X.
SponsorsThis research was supported in part by award KUK-C1-013-04, made by King Abdullah University of Science and Technology (KAUST).Sandia National Laboratories is a multiprogram laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000.