Description
In photometric observations the light of galaxies is observed through a reduced number of filters. This lack of information translates into uncertain estimates in the redshift of said galaxies. This uncertainty is one of leading contributors to the error budget of Stage-IV surveys meaning that it needs to be treated with care. In this talk I will present current efforts to incorporate these uncertainties into the LSST pipeline from individual sources of galaxies all the way to cosmological constraints. I will present a comparison of different uncertainty models from their performance at capturing the measured distribution of galaxies to their final impact in the inferred cosmological parameters in the context of 3x2-pt analyses. I will also discuss the feasibility of marginalizing over the parameters of these models analytically, significantly speeding up the cosmological inference. All of the methods described are already publicly available through the DESC library nz_prior.
