Sequential design of computer experiments Seminar

Abstract
In this talk, we show various strategies for the calibration and emulation of simulators having uncertain inputs and internal parameters, with applications to Computational Fluid Dynamics, tsunami wave modelling and climate models. We provide enhancements of a sequential design approach to build Gaussian Process emulators, where points are chosen according to the entropy-based mutual information criterion. Points are selected according to updated parameters of the covariance structure to boost efficiency. We prove that this method is close to optimal. We illustrate the benefits of our improvements on several synthetic data sets and on the investigation of gravity waves parametrizations in the Whole Atmosphere Community Climate Model.