Karlsruhe 2024 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 73: Flavour physics 3

T 73.2: Talk

Wednesday, March 6, 2024, 16:15–16:30, Geb. 30.41: HS 4

Search for B⁺→ K^*+νν using Lorentz Equivariant Neural Networks at the Belle II Experiment — •Caspar Schmitt, Nikolai Hartmann, Sviatoslav Bilokin, and Thomas Kuhr — Ludwig-Maximilians-Universität München

Searches for rare B meson decays allow precision standard model tests owing to particularly precise theory predictions. At the Belle II experiment, B meson pairs are produced exclusively at a known center of mass energy. This allows searches for decays with invisible final state particles, by fully reconstructing one B meson and thereby kinematically constraining the other.

Current precision of searches for B→ K^*+νν is statistically limited, therefore, reconstruction efficiency and background suppression are key figures. In a novel method, the accompanying B meson decay is inclusively reconstructed and neural networks are used to suppress background contaminations. We demonstrate the feasibility of Lorentz-equivariant graph neural nets in inclusive reconstruction, which respect the physical symmetries of the input features and improve the experimental upper limit on the branching fraction.

Keywords: Belle II; Rare Decay; Machine Learning; Lorentz Equivariance; New Physics