Würzburg 2018 – wissenschaftliches Programm

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

AKPIK: Arbeitskreis Physik, moderne Informationstechnologie und Künstliche Intelligenz

AKPIK 1: Arbeitskreis Physik, IT & KI (AKPIK) I

AKPIK 1.3: Vortrag

Montag, 19. März 2018, 17:15–17:30, 70 - HS 00.107

FACT -- Improving the Angular Resolution of FACT-Tools Using Machine Learning — •Maximilian Nöthe and Kai Arno Brügge for the FACT collaboration — Exp. Physik 5b, TU Dortmund, Otto-Hahn-Str. 4a, 44227 Dortmund, Deutschland

The First G-APD Cherenkov Telescope (FACT) is pioneering the usage of solid state photo sensors (G-APDs aka SiPMs) for groundbased gamma-ray astronomy. FACT is located on the Roque de los Muchachos on the canary Island of La Palma. Since October 2011, the FACT collaboration has successfully been showing the reliability of SiPM for the IACT technique and blazar monitoring.

One of the most challenging tasks in Imaging Air Cherenkov Telescopes is the reconstruction of the origin of incoming particles. Based on the recorded image of the Cherenkov photons, the origin of the gamma ray has to be estimated. This is an especially hard task for single telescopes.

In this contribution, the improvements of the angular resolution and sensitivity of the FACT-Tools analysis are presented, which were gained by using a machine learning based implementation of the so called disp method.

For this approach, two random forests have been trained. First, a random forest regressor to estimate the absolute value and second a random forest classifier to perform the head/tail-disambiguation of the shower.

The results shown were achieved on the publicly available FACT raw data using open source software.