Erlangen 2026 – wissenschaftliches Programm

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

T: Fachverband Teilchenphysik

T 28: Data, AI, Computing, Electronics III

T 28.4: Vortrag

Dienstag, 17. März 2026, 17:00–17:15, KH 00.024

Graph Neural Network based inclusive flavour tagger at the LHCb experiment — •Yukai Zhao¹, Sara Celani², Stephanie Hansmann-Menzemer¹, and Pelian Li³ — ¹Physikalisches Institut, Ruprecht-Karls-Universität Heidelberg, Germany — ²CERN, Switzerland — ³University of Chinese Academy of Sciences, China

The study of CP violation at the LHCb experiment is essential for understanding the observed matter*antimatter asymmetry in the universe. A key component of many such measurements is the decay-time-dependent analysis of oscillating neutral B mesons, which requires knowledge of the b-hadron flavour at production. This initial flavour information can not be determined directly from the decay products of the signal candidate. Instead, it is inferred using flavour tagging algorithms which exploit correlations with other particles produced in the same proton*proton collision. This talk presents a novel flavour-tagging method based on Graph Neural Networks (GNNs). The approach leverages the Deep Full Event Interpretation framework, which performs inclusive reconstruction of heavy hadrons in the event. By modelling the relationships between particles identified through inclusive reconstruction, the GNN-based tagger is expected to enhance the usage of kinematic and topological information, leading to a significant improvement in the estimated flavour-tagging performance. The results represent a promising advancement for time-dependent CP violation measurements at LHCb.

Keywords: Flavour Tagging; LHCb; Graph Neural Networks; Flavour Physics