A new top jet tagging algorithm for highly boosted top jets

A new top-jet-tagging algorithm is presented based on work in Ref [1]. The full results can be seen in Refs [2, 3]. This algorithm uses the Cambridge-Aachen jet clustering algorithm to decompose highly boosted top jets into subjet components and examine the kinematics of these subjets. With this algorithm, an efficiency of 46% for top-jets with p_T = 600 GeV/c is obtained, together with a rejection of 98.5% for non-top jets with p_T = 600 GeV/c. With 100 pb^-1 of 10 TeV pp collider data collected at CMS, narrow tt resonances in the all-hadronic channel, with masses of 1, 2, 3, and 4 TeV/c² can be excluded at the 95% C.L. for cross sections of 17.2, 1.5, 0.7, and 0.8 pb, respectively. A 5σ discovery can be made for cross sections of 43.6, 4.0, 1.6, and 1.3 pb, respectively.