Incorporating Bayesian approaches into modelling astrophysical data
Stefano Andreon1, Merrilee Hurn2
1INAF-Osservatorio Astronomico di Brera, Milano, Italy; 2Department of Mathematical Sciences, University of Bath, Bath, United Kingdom

Two situations often occur in astronomy (and physics):

a) we have information about the things we are measuring before actually measuring them. For example, astronomers regularly use prior information about the type of results they expect to see in order to select the telescope to use, its instrumental set up, and the lenght of the exposure. But this prior knowledge is then often discarded by many astronomers in formulating the model used to analyse the data, resulting in a suboptimal use of the available information, sometimes introducing biases.

b) the quantity of real interest cannot be measured directly but requires the use of noisy proxy measurements which in order to be collected, require in turn some knowledge of the quantity being measured or the value of a related but unknown quantity. In real astronomical experiments, the situation can be tricky and involved. Therefore, very often, astronomers are forced to make assumptions and approximations to reach their goal, a fully-comprehensive and coherent analysis of the data would be too astronomer-time expensive to be set in the frequentist framework.

This talk illustrates the above two situations with real astronomical problems that have been successfully addressed in a Bayesian framework by the authors.

Biography: Stefano Andreon is an astronomer, working at the Osservatorio Astronomico di Brera, in Milano, Italy. He discovered the most distant cluster of galaxies known, at some 10.6 billion light years away, seen when the Universe was only one quarter of its current age. He is author of more than 60 referred papers on major astronomical journals, two third of which as first author, and co-author of the book “Bayesian Methods in Cosmology”.