Speaker
Description
The muon puzzle remains one of the intriguing mysteries in particle physics. To fully understand its origin, we need precise knowledge of the mass composition of ultra-high-energy cosmic rays (UHECRs). At energies above 300 PeV the direct detection of UHECRs is not feasible, necessitating the use of mass-sensitive observables of extended air showers (EAS) induced by UHECRs interacting with the atmosphere. One way to achieve high statistics for these mass-sensitive observables is the use of ground-based detector arrays, such as the Surface Detector (SD) of the Pierre Auger Observatory. The SD consists of three sub-arrays of independent detector stations arranged in triangular grids of different spacing. When an EAS triggers the SD, a subset of stations records the particles of the shower cascade reaching the ground level. Recently, it has been shown that neural networks (NNs) can extract mass-sensitive observables from data taken by the SD-1500, the largest sub-detector of the SD. In this contribution, we demonstrate the feasibility of an NN-based approach to reconstruct high-level shower observables from data simulated for the SD-750, a smaller detector array nested within the SD-1500. A key advantage of the SD-750 is that it contains the Underground Muon Detector of the Observatory, which allows the output of the NNs to be calibrated for mass-sensitive observables, such as the muon number, overcoming the shortcomings of the current generation of air shower simulations. We show, that even though the footprints are not necessarily contained in the SD-750, the training of the NNs remains stable. Moreover, we show that the NN-based approach achieves performance comparable to that of the SD-1500 network.
| Collaboration(s) | Pierre Auger Collaboration |
|---|
