The Planck Likelihood analysis

The Planck Likelihood is one of the main cosmological products of the Planck mission. It compresses the Planck data into a set of statistical descriptions of our best full sky determination of the CMB temperature and polarization, and constitutes an important element of the Planck Legacy. Until a new satellite mission, with both improved resolution and sensitivity, can provide us with a new full sky determination of the CMB, some of the scales probed by the Planck data will remain a reference for all CMB observations.Thanks to numerous improvement at all steps of the Planck data processing, the legacy likelihood represents a solid improvement over the 2015 Planck Likelihood, both at large scales (with a greatly improved constraint on the large scale polarization) and at small scales (with a solid leakage model, allowing to recommend the use of the polarized data for cosmology). I will review the determination and validation of the Planck Likelihood. I will describe the different choices made to build at each of the scales probed by Planck a good approximation of the likelihood. I will explain how we modelled and mitigated the different non-idealities of the data (e.g. astrophysical foregrounds, instrumental effects...), describe the different validation we performed, based on comparing different data cuts, or using extensive End-to-End simulations, and finally give an estimate of the level of possible residual systematics.