Adaptive Stochastic Collocation on Sparse Grids



We consider Partial Differential Equations (PDEs) with random parameters, solved by adaptive stochastic collocation. Our research studies common error indicators compared to adjoint based error estimates extended to the stochastic framework. We show that such estimates are able to detect the true error very accurately and suggest to replace the expensive adjoint problem by a reduced model. (© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)