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

T: Fachverband Teilchenphysik

T 18: Methods in Particle Physics I (Calo, Jets, Tagging)

T 18.1: Vortrag

Montag, 31. März 2025, 16:45–17:00, VG 4.101

Calibration of calorimeter signals in theATLAS experiment using an uncertainty-awareneural network — •Isabel Sainz Saenz-Diez — Kirchhoff Institute for Physics, Heidelberg University

Measuring energy deposits in the calorimeters are a key aspect of particle reconstruction. In the case of the ATLAS experiment at the Large Hadron Collider (LHC), the calorimeter signals are reconstructed as clusters of topologically connected cells (topo-clusters). These are calibrated in such way that they correctly measure the energy deposited by electromagnetic showers, but they do not compensate for the fraction of energy that does not contribute to the signal, which is part of the hadronic showers. In order to account for this energy, a local hadronic calibration of topo-clusters is applied. Machine Learning (ML) methods have been proposed as an alternative to the current hadronic calibration in ATLAS. Both a Deep Neural Net (DNN) and a Bayesian Neural Net (BNN) yield continuous unbinned calibration functions with an improved performance with respect to the standard calibration. Additionally, the BNN provide an estimation on the uncertainties of the calibration output. The talk will present the current status of the implementation and performance of the proposed models.