Speaker
Description
We develop a machine learning algorithm that infers the radial profiles of total and gas mass of galaxy clusters given thermal Sunyaev-Zeldovich (SZ) effect maps. The architecture is composed of a combination of an autoencoder and a random forest. The first is used to extract the information from the maps, while the second performs the final estimation of the radial mass profiles. This ML algorithm is trained and tested on a sample of 73,138 mock SZ maps. Each map is generated from one of 29 projections of 2522 galaxy clusters from The Three Hundred simulation. We show that the model can reconstruct the gas mass profile, responsible for the SZ effect, but also the total mass one, without any a priori assumption on the physics of the cluster. We demonstrate that both total and gas mass radial profiles are unbiased with a scatter of about 10% slightly increasing towards the core and the outskirt of the cluster.
We selected the clusters in a wide mass range, 10^13.5 - 10^15.5 h^-1 Msun and in a different dynamical state from very relaxed to very disturbed. We see that both the accuracy and precision of this method have a slight dependence on the dynamical state, but not on the cluster mass.
In summary, we find ML techniques to offer a powerful method to predict model-independent mass profiles for large samples of clusters.
