Bayesian asteroseismology

Abstract

This thesis presents a new probabilistic method for the asteroseismic analysis of stellar structure and evolution with the goal of providing a universal tool to improve our knowledge of stellar modelling. This new method implements the advantages of Bayesian analysis, such as the treatment of systematic errors and nuisance parameters, the modular structure of Bayesian analysis, and the correct normalization of all probabilities. First, a general introduction to asteroseismology is provided, followed by an comprehensive guide to Bayesian analysis. The derivation of the new method then follows, and its subsequent application to current problems in asteroseismology is also presented. An in-depth analysis of the Sun is performed in order to investigate long standing problems with the solar chemical composition. This also reveals the presence of systematic problems in the modelling of the Sun, potentially requiring new developments in solar modelling. Finally, the new method is also applied to 23 stars that were observed with the Kepler satellite, in order to perform a comparative investigation with respect to published results from other teams, and to study systematic errors in the stellar models.