43rd International Conference on High Energy Physics

Calibration of GN2 jet-tagging algorithm using t¯t events from partial Run 3 pp collision data collected with the ATLAS detector

Not scheduled

20m

Natal, Brazil

Poster (early career: student or postdoctoral researcher) Operation, Performance and Upgrade of Present Detectors

Speakers

Irena Nikolic (LPNHE Université Paris Cité CNRS) Reisaburo Tanaka (IJCLab, Université Paris-Saclay (FR))

Description

Talk Abstract Algorithms designed to identify the flavor of jets are an essential component of the ATLAS physics program, enabling the observation of Higgs boson decays to bottom quarks and high-precision measurements of the top quark. For the LHC Run 3 data-taking period, ATLAS utilizes GN2, a transformer-based, state-of-the-art jet-tagging algorithm. GN2 provides a significant improvement in the discrimination of b-jets from c-jets and light-flavor jets. To ensure the algorithm’s reliability, the predicted efficiency in simulation is compared with the efficiency measured in data, to calculate corrective scale factors (SFs). This calibration is performed using t¯t events across two channels: The dileptonic channel, which provides a high-purity sample of b-jets, is used for b-jet calibration, while the semi-leptonic channel is used to simultaneously calibrate both the b-jet tagging and c-jet mis-tagging efficiencies. These results are obtained from the partial Run 3 dataset corresponding to an integrated luminosity of 56 fb−1 taken at a center-of-mass energy of √s = 13.6 TeV.