Title: Modeling and statistical separation of components in astrophysics
Abstract: One of the challenges of astrophysics and cosmology is to study complex non-linear processes from a limited number of multi-component observations. This task is made even more difficult by the fact that the physical modeling of these processes is not always complete, which often implies to rely only on available observations, without any prior training step. In this seminar, we will study how to build efficient low-dimensional models taking into account the physical character and regularity of the processes studied. These maximum entropy models, built from scattering transforms, can be constructed directly from observational data. We will then discuss how these tools allow the development of new types of component separations, allowing in particular to estimate the statistics, and thus to build a model, of unknown processes from multi-component observations.