The presentation will focus on the calibration of computer code for the purpose of studying a physical phenomenon or system that it models.
By calibration we mean the estimation of the parameters of the computer code, which is expensive in terms of execution time.
We begin with an introduction to the statistical framework, followed by an exposition of Bayesian statistical
tools such as MCMC sampling methods, Gaussian process modeling and others.
Criteria for selecting physical and numerical experiments for calibration will then be discussed.
Illustrative examples will punctuate the presentation.
Finally, an application to an analytical toy case will be presented.
BE CAREFUL : this presentation will be in FRENCH
