Lunch 12:30pm; Talk 1pm
How should we analyse the data from an experiment or observation? The ideal Bayesian answer to this is to calculate the posterior probability of the quantities that we wish to infer. In cosmology, this are typically the cosmological parameters. The posterior is not generally analytic, but one can attempt to draw samples from the posterior, and if sufficient samples can be drawn, then the job is done - the set characterises essentially everything we have learned. Bayesian Hierarchical Models (BHMs) are powerful tools that can allow a principled analysis of data, properly propagating errors from each stage. Recent theoretical advances have allowed this approach to be applied to weak lensing data, simultaneously inferring statistical properties of the cosmic shear field and the true shear fields, a parameter space that has hundreds of thousands of dimensions. The approach also allows some of the most difficult problems in weak lensing, such as a complicated sky mask and intrinsic alignments, to be easily and naturally included. In this talk, I will show how the method works, and show preliminary results from analysis of the CFHTLenS survey. As a general technique, BHMs can be applied in many analysis problems.
Alan Heavens studied as an undergraduate and postgraduate at the University of Cambridge, UK, and then moved to Edinburgh. He was appointed as Director of the new Imperial Centre for Inference and Cosmology, at Imperial College London, in 2012. His research interests cover many areas of cosmology, including weak gravitational lensing, large scale structure, and the cosmic microwave background. He is also a director of a spin-out company applying inference solutions to medical imaging.