Aachen 2019 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 29: Deep Learning II

T 29.9: Talk

Tuesday, March 26, 2019, 18:00–18:15, H06

Deep Learning based Air Shower Reconstruction at the Pierre Auger Observatory — •Jonas Glombitza, Martin Erdmann, Maximilian Vieweg, and Michael Dohmen — III. Physikalisches Institut A, RWTH Aachen

The surface detector of the Pierre Auger Observatory measures the footprint of ultra-high energy cosmic ray induced air showers on ground level. Furthermore, fluorescence telescopes allow hybrid detection of air showers and hence, for an independent crosscheck. Reconstructing observables sensitive to the cosmic ray mass, is a challenging task and mainly based on the fluorescence detector which, however, has a small duty cycle. Recently, great progress has been made in multiple fields of machine learning by using deep neural networks and associated techniques. Applying these new techniques on air shower physics provides a new and independent reconstruction.

In this talk, we present AixNet, a deep convolutional neural network for the reconstruction of ultra-high energy cosmic rays properties [1]. First, we assess the performance on CORSIKA based air showers, discuss the performance limit and compare the performance achieved on data by cross-calibrating with the fluorescence detector. Furthermore, we visualize the multidimensional differences between data and simulation using deep neural networks to understand differences in the reconstruction and prepare the simulation for refinement studies [2].

[1] DOI: 10.1016/j.astropartphys.2017.10.006

[2] DOI: 10.1007/s41781-018-0008-x