Bayesian emulators and the stochastic finite element method

Diaz De la O, F. A. and Adhikari, S.
The Ninth International Conference on Computational Structures Technology, Athens, Greece, September 2008.

A method for efficiently representing realizations of computationally expensive random fields is proposed. The approach, known as the emulator, consists of building a statistical approximation of realizations of such random fields. It is based on few runs of a code that performs the discretization of the random field via the Karhunen-Lo\`{e}ve expansion. This study is aimed at the incorporating emulators to reduce the computational cost of stochastic finite element codes. Numerical results of emulating realizations of a Gaussian homogeneous two-dimensional random field are presented. It is shown that proposed approach can reduce simulation cost of random fields for stochastic finite element applications.
BiBTeX Entry
@INPROCEEDINGS{cp37,
    AUTHOR={F. A. DiazDeLaO and S. Adhikari},
    TITLE={Bayesian emulators and the stochastic finite element method},
    BOOKTITLE={The Ninth International Conference on Computational Structures Technology},
    YEAR={2008},
    Address={Athens, Greece},
    Month={September},
    Note={}
}

by Sondipon Adhikari