New graph-neural-network flavor tagger for Belle II and measurement of sin 2β’π 1 in π΅ 0 β π½/πβ’πΎ$^0_ S$ decays
We present GFlaT, a new algorithm that uses a graph-neural-network to determine the flavor of neutral π΅ mesons produced in Οβ‘(4β’π) decays. It improves previous algorithms by using the information from all charged final-state particles and the relations between them. We evaluate its performance using π΅ decays to flavor-specific hadronic final states reconstructed in a 362 fb β1 sample of electron-positron collisions collected at the Οβ‘(4β’π) resonance with the Belle II detector at the SuperKEKB collider. We achieve an effective tagging efficiency of (37.40 Β± 0.43 Β± 0.36%), where the first uncertainty is statistical and the second systematic, which is 18% better than the previous Belle II algorithm. Demonstrating the algorithm, we use π΅ 0 βπ½/πβ’πΎ$^0_ S$ decays to measure the mixing-induced and direct πΆβ’π violation parameters, π = (0.724 Β± 0.035 Β± 0.009) and πΆ = (β0.035 Β± 0.026 Β± 0.029).