Recent Developments in Stochastic Reduced Basis Projection Schemes

Sachin K. Sachdeva, Prasanth B. Nair, Andy J. Keane46th 2005 · DOI

Abstract

In this paper we present an overview of recent developments in stochastic reduced basis projection schemes with an emphasis on applications to stochastic finite element analysis. We begin with an outline of the basic steps involved in discretization of stochastic partial dierential equations to obtain a system of linear random algebraic equations. Subsequently, we present the standard formulation that uses basis vectors spanning the preconditioned stochastic Krylov subspace to represent the solution process. Next, we present a hybrid formulation which combines stochastic reduced basis methods with polynomial chaos expansions to tackle non-Gaussian uncertainty models. Finally, we present another stochastic projection scheme referred to as the strong Galerkin projection scheme which imposes orthogonality in a stricter sense. The strong Gakerkin scheme is presented as a generalization of stochastic reduced basis methods based on the standard weak Galerkin condition. To illustrate the performances of various projection schemes we present a selection of results obtained by analysing a linear structural problem subjected to static loads when the Young’s modulus is modeled by Gaussian as well as lognormal random fields. Our numerical studies suggest that the strong Galerkin projection scheme gives better results compared to the standard weak Galerkin scheme for the same number of basis vectors chosen from the stochastic Krylov subspace.
BibTeX
@inproceedings{sachdeva2005recent,
  author = {Sachin K. Sachdeva and Prasanth B. Nair and Andy J. Keane},
  title = {{Recent Developments in Stochastic Reduced Basis Projection Schemes}},
  booktitle = {46th},
  year = {2005},
  doi = {10.2514/6.2005-1950},
}

Loading related content...

Figure