Heidelberg 2022 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
T: Fachverband Teilchenphysik
T 43: DAQ and Trigger 2
T 43.5: Vortrag
Dienstag, 22. März 2022, 17:15–17:30, T-H28
Optimizing the ATLAS b-jet Trigger for the LHC Run 3 — •Victor H. Ruelas Rivera — Humboldt-Universität zu Berlin, Berlin, Germany
The Higgs potential provides a way to experimentally probe and understand the underlying principles of mass generation and electroweak symmetry breaking. The shape of the Higgs potential is proportional to the Higgs self-coupling, λHHH, which can be probed at the LHC via proton-proton collisions (pp → HH). Di-Higgs to 2 pairs of quarks (HH → bbbb) is one of the most sensitive decay channels and it relies heavily on b-jet triggers. Triggers select information in real-time from the collisions and help mitigate ATLAS data acquisition getting overwhelmed by QCD jets. However, HH → bbbb is difficult to trigger due to the soft signal kinematics and high thresholds of hadronic triggers. Hence, more signal can be gained in Run 3 with better b-jet triggers. One of the goals of Run 3 is to improve the HH → bbbb triggers to enhance signal acceptance of SM and Beyond Standard Model (BSM) scenarios. The taggers that will be used for the b-jet trigger in Run 3 exploit multivariate analysis techniques, mainly Deep Neural Networks. This talk presents the optimization of the neural-network based flavour tagging discriminant used by the b-jet trigger. The algorithm is optimized on tracks, jets and vertices reconstructed by the High Level Trigger (HLT) software. The training software, network architecture and simulated events are being shared with the ATLAS offline b-tagging group to address redundancies and combine efforts towards a unified training framework for quick model re-optimization.