Reconstructing UHECR Primaries from EAS Characteristics using an ML Model

Millions of ultra-high-energy cosmic rays (UHECRs) strike our atmosphere every day. UHECR primaries are typically atomic nuclei — thought to be somewhere between a bare proton and iron — which are accelerated by energetic cosmic events to extremely high per-particle energies (on the order of 10^18 eV). Upon reaching Earth, these UHECRs interact to produce cascades of particles called Extensive Air Showers (EAS) that emit fluorescent light as they move through the atmosphere. At the Pierre Auger Observatory (PAO), fluorescence telescopes can observe these light emissions to reconstruct properties of the primary. The mass of the primary is a particularly important quantity and has proven difficult to reconstruct with current methods. In a recent paper, our group developed a machine learning (ML) approach which shows the potential to provide a higher mass-sensitivity than so far described. ML models are useful for this task because these algorithms have the potential to discover subtle EAS features that are related to the mass of the primary. The current model architecture was designed to be trained on noised EPOS-LHC and Sybill 2.3d simulations. The goal of the project described in this contribution is adapting this method to reconstruct the mass of events observed at the PAO. This contribution will report on the progress and application of this model to realistic simulated PAO fluorescence detection events and outline next steps.