Robustness of Planck results to statistical methods: the case of neutrinos

International audienceWith its wide frequency coverage Planck is a unique tool not only to test the standard cosmo- logical model (LCDM) but to explore beyond it, looking, for example, for the absolute scale of neutrino masses. Cosmological parameters are traditionally determined using a Bayesian approach based on Markov chains (MCMC) but a Frequentist approach based on profiles- Likelihood is also possible. The two methods are presented, with special focus on the latter, which gives results remarka- bly consistent with Bayesian approach for the LCDM model. The interesting case of neutrinos is discussed. Here the presence of a physical boundary on the masses induces differences in the two approaches nevertheless giving results in perfect agreement