Speaker
Description
Recent measurements of cosmological parameters highlight possible tensions between values determined by CMB measurements and those determined by late epoch observations which could potentially challenge the current theoretical model, in particular for the clumpiness parameter σ8 which is related to the normalization of the primordial matter power spectrum. In this project, we investigate this tension by constructing a forward cosmological modeling pipeline to determine this parameter from galaxy cluster counts using simulation based inference. The implemented pipeline generates samples of galaxy clusters from the halo mass function and includes all the relevant observational effects, which allows us to apply the exact selection function of a survey. We go beyond examining the redshift and flux distributions of clusters by incorporating as well their temperature distribution, which allows to some extent to break the degeneracy between Ωm and σ8. We then train a convolutional neural network to learn the mapping between model parameters and observed data. Finally, we apply our framework to the measured distributions of redshifts, fluxes, and temperatures in the XXL survey to obtain the parameters that best reproduce the observed sample. Our simulation based inference framework can be readily applied to the new large-scale galaxy clusters surveys such as eROSITA, Euclid and SPT-3G.
