Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

T: Fachverband Teilchenphysik

T 38: Data analysis, information technology II

T 38.2: Vortrag

Dienstag, 16. März 2021, 16:15–16:30, Tm

Deep-Learning-Based Reconstruction of Cosmic-Ray Properties From Extensive Air Shower MeasurementsMartin Erdmann, Jonas Glombitza, Berenika Idaszek, and •Niklas Langner — III. Physikalisches Institut A, RWTH Aachen University

Ultra-high-energy cosmic rays colliding with the Earth's atmosphere lead to the formation of extensive air showers. At the Pierre Auger Observatory, showers are measured using the surface detector (SD) on ground and the fluorescence detector (FD) observing the sky above. The depth of shower maximum, directly measurable by the FD, is of particular interest due to its connection to the cosmic-ray mass.

Currently thriving in the field of machine learning, deep neural networks which consist of hundreds of thousands of parameters can be trained to exploit complex data and extract information otherwise hard to access. While such networks are able to achieve remarkable precision, understanding their working principle is challenging due to the large number of parameters. Using deep learning, the depth of shower maximum can be extracted from SD observations.

We present our network to extract properties of air showers by analyzing the signal of water Cherenkov detectors. It utilizes recurrent long short-term memory layers and hexagonal convolutions. The technical setup and method is explained. We investigate the reasoning of the trained network by visualizing inputs relevant for the network's decision. We show the performance of our method using simulations and discuss the incorporation of additional scintillator detectors which are part of the upgrade program of the Pierre Auger Observatory.

100% | Bildschirmansicht | English Version | Kontakt/Impressum/Datenschutz
DPG-Physik > DPG-Verhandlungen > 2021 > Dortmund