Boosted X->bb tagger calibration using semi-leptonic ttbar events collected with the ATLAS detector

Flavor-tagging is a critical component of the ATLAS experiment physics program, as many precision measurements and searches for new physics depend on the accurate identification of jets containing b-hadrons (b-tagging). A novel b-tagging approach uses a single machine learning model based on a transformer architecture to process information from a variable number of tracks and other objects in the jet in order to simultaneously predict the jets flavor, the partitioning of tracks into vertices, and the physical origin of each track. Notably, the versatility of the approach is demonstrated by its successful application in boosted Higgs tagging using large-R jets. To extract the mistagging efficiency of boosted t¯t large-R jets, an in-situ calibration is performed using semi-leptonic t¯t events. The data to simulation scale factors are derived using the Run 2 pp collision data collected by ATLAS experiment at √s = 13 TeV, with the integrated luminosity of 139 fb−1.