Measurement of the $b$-jet identification efficiency in dileptonic $t\bar{t}$ events using proton-proton collision data at $\sqrt{s}=13.6$ TeV collected with the ATLAS detector

This paper presents the performance of the identification of jets containing $b$-hadrons ($b$-jets) for the GN2 algorithm, a transformer-based model for jet flavour tagging, using data collected by the ATLAS detector at the LHC. The analysis uses proton-proton collision data recorded in 2022 and 2023 at a centre-of-mass energy of $\sqrt{s}=13.6$ TeV, corresponding to an integrated luminosity of 56 fb${}^{-1}$. The $b$-jet identification efficiency and jet flavour composition are extracted simultaneously from a sample enriched in top-quark pair events ($t\bar{t}$). This efficiency is measured as a function of the jet transverse momentum in the range of 20-400 GeV and across six intervals of cumulative efficiency as derived in simulated $t\bar{t}$ events: [100%, 90%], [90%, 85%], [85%, 77%], [77%, 70%], [70%, 65%], and [65%, 0%]. The GN2 algorithm demonstrates significant performance improvements over its predecessor DL1d, achieving up to a factor of two (three) higher rejection of light-flavour (charm-flavour) jets at the same $b$-jet efficiency. The measured efficiencies in data are compared with simulation to derive correction factors ranging from 0.9 to 1.3. The total uncertainty is around 1% for jets with transverse momentum larger than 60 GeV in the [65%, 0%] interval of cumulative efficiency.