LHCb Turbo++ Slimming: Inclusive B Hadron Decay Product Selection

My summer student project focused on reducing the turbo++ output bandwidth for B meson decays at LHCb. The aim is to discard as many of the background tracks as possible, while still retaining almost all of the signal events. I analysed the ability to distinguish signal from background tracks, by considering four different B meson decays, while using signal definitions for just one particular decay, as well as for a general type of decay. I added tracks to a triggered turbo candidate, to reconstruct the B meson. I used a boosted decision tree and an artificial neural network to classify events, and compared the resulting ROC curves.