EeV photons from the Pierre Auger Observatory: a Real-time probe for Multi-messenger analyses

The quest for the origin of cosmic rays intrinsically implies a multi-messenger approach. Due to magnetic fields that permeate the universe, cosmic rays,  which are mostly charged ions, do not point back to the sources.  Direct information about their acceleration sites can, however, be obtained by searching for the neutral particles, gamma-rays and neutrinos,  that can be generated by the interactions of cosmic rays at the acceleration sites. In this work, the search for nearby ultra-high-energy transient sources is addressed by performing a search for photon candidates among the events collected by the Pierre Auger Observatory.  The goal, in the future, is to perform a real-time coincidence search with neutrinos collected by the IceCube Observatory.

Photon-induced air shower are characterized by a deeper depth at the shower maximum and a lower number of muons with respect to the bulk of hadron-induced background. Proxies for these parameters can be measured by the Pierre Auger Observatory, which is composed by a fluorescence detector (FD) and a ground array of particle detectors (SD). In this work we present a new technique based on the use of a Deep Neural Network (DNN) to identify photon-like events with energies above 1 EeV. The DNN is trained by using the signals and time information provided by the SD detector. The approach is tested on full air shower simulations, and is explored in terms of photon/hadron separation capability and reconstruction accuracy.