Muon measurements at the Pierre Auger Observatory

TL;DR

This paper reviews muon measurements at the Pierre Auger Observatory, discusses the muon deficit problem in air shower simulations, and introduces a neural network method to better extract muon signals for improved cosmic ray analysis.

Contribution

It presents new insights into muon fluctuations, addresses the muon deficit issue, and develops a neural network approach for muon signal extraction in surface detectors.

Findings

Muon simulations underestimate muon numbers compared to data.

Muon fluctuations provide new insights into hadronic interaction models.

Recurrent neural networks can effectively extract muon signals from detector data.

Abstract

The Pierre Auger Observatory is the world's largest detector for observation of ultra-high-energy cosmic rays (UHECRs) (above the energy of $10^{17}$ eV). It consists of a Fluorescence Detector (FD) and an array of particle detectors known as the Surface Detector (SD). Observations of extensive air showers by the Observatory can be used to probe hadronic interactions at high energy, in a kinematic and energy region inaccessible to experiments at man-made accelerators and to measure the muon component of the shower. Air showers induced by different primaries have different muon contents. With increasing mass of the primary cosmic ray particle, it is expected that the muon content in the corresponding air showers should also increase. Recent results obtained from the Pierre Auger Observatory and other experiments indicate that all the shower simulations underestimate the number of muons in the showers compared to the data. This is the so-called muon deficit. In this paper we briefly review the muon measurements, and present in more detail recent results on fluctuations in the muon number. These results provide new insights into the origin of the muon deficit in air shower simulations and constrain the models of hadronic interactions at ultrahigh energies. With the current design of the surface detectors it is also difficult to reliably separate the contributions of muons to the SD signal from the contributions of photons, electrons, and positrons. Therefore, we also present a new method to extract the muon component of the signal time traces recorded by each SD station using recurrent neural networks. The combination of such algorithms, with the future data collected by the upgraded Pierre Auger Observatory, will be a major step forward, as we are likely to achieve an unprecedented resolution in mass estimation on an event-by-event basis.