Algorithms to identify high-energy B hadrons via their hit multiplicity increase through pixel detection layers
Manuel Sommerhalder
Prof. Dr. Ben Kilminster
Dr. Thea Årrestad
Dr. Yuta Takahashi
Track-based b tagging algorithms are shown to suffer significant efficiency loss for jets with a transverse momentum above 1 TeV due to track reconstruction inefficiencies. In the same high-momentum region, a simple hit counting approach performs similarly or even exceeds the performance of the most commonly used b tagging algorithm in the CMS experiment, the Combined Secondary Vertex Tagger (CSV). Different implementations of cut-based discriminants, as well as neural network models are shown to yield a relative efficiency gain of up to 112% for jets with transverse momenta above 1.2 TeV for a mistag rate of 10% when used in addition to the CSV.