The Three Hundred project: A Machine Learning method to infer clusters of galaxies mass radial profiles from mock Sunyaev-Zel'dovich maps

Abstract

We develop a machine learning algorithm to infer the 3D cumulative radial profiles of total and gas mass in galaxy clusters from thermal Sunyaev-Zel'dovich effect (SZ) maps. By using 2,522 simulated clusters from the T T H project at redshift 𝑧 < 0.12, we generate more than 73,000 mock images along several lines of sight and train a model that is a combination of an autoencoder and a random forest. The model is able to reconstruct the 3D gas mass profile, responsible for the SZ effect, but also the total mass one, including the contribution of the dark matter component, without any a priori assumption on the physics of the clusters. We show that the recovered total and gas mass radial profiles are unbiased with a scatter of about 10%, slightly increasing towards the core and the outskirts of the cluster. We selected clusters in a wide mass range, 10 13.5 ≤ 𝑀 200 /( ℎ −1 M ) ≤ 10 15.5 and spanning different dynamical states, from very relaxed to very disturbed halos. We verify that both the accuracy and precision of this method show a slight dependence on the dynamical state, but not on the cluster mass. To further check the consistency of our model, we fit the inferred total mass profiles with a NFW analytical model and compare the resulting concentration values with those of the true profiles. We observe that for higher values of the concentration, the inferred profiles are estimated without bias, reproducing a reliable mass-concentration relation. Finally, the comparison with a more standard mass estimate method, such as the hydrostatic equilibrium, shows that our technique recovers the total mass that is not affected by bias due to the non thermal motions of the gas.