The role of model selection, summary statistics and multiple-point statistics in Bayesian inverse problems: Examples from geophysics.

Lecturer: Niklas Linde (UNIL)

Abstract

Model selection and multiple-point statistics allow for comparisons of alternative conceptual descriptions of the system under study, while summary statistics (Approximate Bayesian Computation) allows us to relax some of the assumptions made in terms of observational, modelling and prior uncertainties. These concepts will be introduced and motivated with examples from geophysics.