Design of a Deep Full Event Interpretation for LHCb and application in semitauonic B decays

ended

The LHCb experiment has shown an excellent performance in precision measurements of beauty- and charm-hadron decays during the LHC runs 1 and 2. Now, it is undertaking its first major upgrade, to face the challenges of a large increase in instantaneous luminosity, that implies augmented event complexity and background levels. This impacts the trigger, which needs to quickly and efficiently identify signals, while facing limits on the disk storage capacity. The proposed project aims at fighting these limitations in a new way for LHCb: performing a Deep Full Event Interpretation (DFEI) during trigger, where a deep neural network processes the low-level information from the detector and infers the heavy-hadron decays that occurred in the event. This allows to quickly identify different types of background and provides enhanced information on the mother particles. As a first application of DFEI, the first angular analysis of semitauonic decays at LHCb is proposed. Semitauonic decays are partially reconstructed in LHCb, due to the presence of neutrinos in the final state. This leads to high levels of backgrounds, hard to separate from the signal, which makes them a perfect validation bench for the power of DFEI. From a physics perspective, the recently-observed flavour anomalies defy the concept of Lepton Universality of the Standard Model (SM), and point towards potential beyond-the-SM effects in semitauonic B decays. The complementary aim of this project is to perform the first LHCb angular analysis of both B0 -> D+ tau- nu and B0 -> D*+ tau- nu decays, where D*+ -> D+ pi0 and tau- -> mu- nu nu. This analysis will provide crucial information to discriminate between different new models proposed to explain the anomalies. The ambitious projects in this proposal are solidly based on my experience with semitauonic decays, angular analyses and the creation of new software tools, as well as the broad experience of the host group in event reconstruction and data analysis.