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

T: Fachverband Teilchenphysik

T 103: Methods in Particle Physics V (Event Reconstruction, PID)

T 103.1: Vortrag

Freitag, 4. April 2025, 09:00–09:15, VG 4.101

Improving Reconstruction in the Belle II Electromagnetic Calorimeter Using Graph Neural Networks — •Jonas Eppelt and Torben Ferber — Karlsruher Institut of Technology

Belle II uses an Electromagnetic Calorimeter (ECL) built from Cesium-Iodide crystals to measure a particle’s energy. The current clustering algorithm faces significant challenges from high background conditions, low momentum minimal ionizing particles, and hadronic particles creating multiple clusters. This affects energy resolutions, detection efficiencies for low energetic photons, and higher-level variables used in many analyses. Graph Neural Network(GNN) based methods can leverage more of the available information from the ECL and better represent the sparse and irregular geometry of the clusters. This talk will present ongoing efforts to reduce background, improve energy resolution, and analyze other variables.

Keywords: Calorimeter,; Machine Learning; Graph Neural Networks; Clustering; Belle~II