Abstract

In principle the statistics of the CMB are simple. The fluctuations are to a very good approximation an isotropic Gaussian random field and the distribution of their multipole coefficients is entirely characterised by the power spectrum. This means that, with access to the true multipoles, using the full information content of the CMB to study cosmology becomes a straightforward task. Unfortunately it is impossible to measure the actual multipole coefficients of the fluctuations. Foreground contamination of the data requires masking of the sky and introduces complicated correlations between the measured multipoles. This turns CMB analysis into a complicated and computationally demanding problem. Likelihood functions cannot be evaluated exactly any longer but feasible estimation methods like the popular pseudo-Cell estimation tend to lose information on cosmology.

I will talk about recent work that aims at avoiding information loss in the analysis of the CMB anisotropies on a masked sky. In particular I will present an approach that extends the standard pseudo-Cell method and gives rise to numerically tractable estimators that recover the lost information and produce estimates that are close to optimal.