New graph-neural-network flavor tagger for Belle II and measurement of sin 2ϕ1 in B0→J/ψKS0 decays

Abstract

We present GFlaT, a new algorithm that uses a graph-neural-network to determine the flavor of neutral B mesons produced in ϒ(4S) 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 B decays to flavor-specific hadronic final states reconstructed in a 362 fb−1 sample of electron-positron collisions collected at the ϒ(4S) resonance with the Belle II detector at the SuperKEKB collider. We achieve an effective tagging efficiency of (37.40±0.43±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 B0→J/ψKS0 decays to measure the mixing-induced and direct CP violation parameters, S=(0.724±0.035±0.009) and C=(−0.035±0.026±0.029). Published by the American Physical Society 2024