Overview of an Approximation Theory for Validation

Roger Ghanem
University of Southern California

I will present an overview of a coherent mathematical framework that permits the formulation of various uncertainty quantification issues that are relevant to verification and validation. This framework permits the description of both aleatory and epistemic uncertainties as well as their propagation through predictive models. It is shown that by suitably enlarging the mathematical context in which deterministic computational science problems are cast, validating a prediction can be formulated as a problem in approximation theory.
I will also discuss outstanding implementation challenges.

Return to Program