Speaker
Description
The Square Kilometre Array is expected to measure the 21-cm signal from the Epoch of Reionization (EoR) in the coming decade, and currently operating radio-interferometers might provide a detection even earlier. Recently reported upper limits, including the latest LOFAR results, have already started constraining the astrophysical parameters of the models of the EoR. The challenges that lie ahead are twofold : correctly interpreting the observed data requires accurate Bayesian inference methods as well as realistic simulated models.
Using Loreli II, a database of hydro-radiative Licorice simulations of the EoR and the 21-cm signal, we show the impact of using several inference methods and discuss the commonly-used Gaussian likelihood approximation. On mock power spectra affected by thermal noise that corresponds to 100h of observations with the SKA, we report average differences between inference methods significantly smaller than the width of the posterior. Conversely, the impact of choices and approximations in the different 21-cm simulation codes used by the EoR community remains poorly understood. We discuss this aspect in light of the SKA Data Challenge 3b, as well as show the preliminary results of a comparison between 3D and 1D radiative transfer algorithms used respectively in Licorice and a version of the Beorn code modified to use the Licorice source model and applied to the Loreli II simulations. We discuss the impact of this modelling uncertainty on inference and on the interpretation of current upper limits.
