Die DPG-Frühjahrstagung in Bonn musste abgesagt werden! Lesen Sie mehr ...

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

T: Fachverband Teilchenphysik

T 5: Machine Learning: QCD and electromagnetic showers

T 5.1: Vortrag

Montag, 30. März 2020, 16:30–16:45, H-HS III

Deep Learning-based Air-Shower Reconstruction at the Pierre Auger Observatory — Martin Erdmann, •Jonas Glombitza, and Alexander Temme for the Pierre Auger collaboration — III. Institut A, RWTH AACHEN UNIVERSITY

Ultra-high energy cosmic rays are the most energetic particles found in nature and originate from extragalactic sources. When propagating within the Earth's atmosphere these particles induce extensive air showers which can be measured by cosmic-ray observatories.

The hybrid design of the Pierre Auger Observatory features a large array of surface-detector stations which is overlooked by fluorescence telescopes. The reconstruction of event-by-event information sensitive to the cosmic-ray mass is a challenging task and so far, mainly based on the fluorescence detector observations with their duty cycle of about 15%.

Recently, deep learning-based algorithms have shown to be extraordinary successful across many domains in computer vision, engineering and science. Applying these algorithms to surface-detector data opens up possibilities for improved reconstructions. In particular it allows for an event-by-event estimation of the cosmic-ray mass, exploiting the 100% duty cycle of the surface detector.

In this contribution we present our deep network, based on recurrent layers and hexagonal convolutions. We show the performance of our method and discuss solutions to systematic biases. Finally, we evaluate the performance by comparing the deep learning-based reconstruction to measurements of the fluorescence detector using Auger hybrid data.