Karlsruhe 2024 – scientific programme

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

T: Fachverband Teilchenphysik

T 36: Gamma astronomy 2

T 36.7: Talk

Tuesday, March 5, 2024, 17:30–17:45, Geb. 30.22: kl. HS A

Providing Uncertainty Predictions for Reconstructed CTA Events Using Neural Networks — •Cyrus Walther and Maximilian Linhoff for the CTA collaboration — TU Dortmund University, Dortmund, Germany

The Cherenkov Telescope Array (CTA) is the next generation of ground-based, gamma-ray astronomy experiments. The multi-telescope arrays in La Palma, Spain, and Paranal, Chile, will outperform the state-of-the-art imaging atmospheric Cherenkov Telescopes (IACTs) by one order of magnitude in sensitivity. In the event analysis of IACTs, the reconstruction of the energy and direction of the primary particle is crucial. Many analysis methods do not provide reliable uncertainty estimations on the reconstructed energy and direction. Based on successful work in the IceCube collaboration, we want to overcome this challenge with a neural network approach. Using a gaussian likelihood loss function, a prediction of uncertainties in energy and direction on CTA simulation data is performed and assessed.

Keywords: Neural Networks; Gamma Astronomy; Uncertainty