000004233 001__ 4233
000004233 005__ 20240408112232.0
000004233 037__ $$aBELLE2-MTHESIS-2024-008
000004233 041__ $$aeng
000004233 100__ $$aLukas Goldschmied
000004233 245__ $$aMultivariate analysis methods for the optimization of searches for new physics with leptons at Belle II
000004233 260__ $$aVienna$$bHEPHY$$c2022
000004233 300__ $$a69
000004233 500__ $$aPresented on 29 04 2022
000004233 502__ $$aMSc$$bVienna, Vienna Univ of Technology$$c2022
000004233 520__ $$aThis thesis will cover the use of multiple multivariate analysis (MVA) methods ap- proaches for the identification of Leptons in the Belle II experiment. First, I will compare two machine learning techniques in terms of classification and performance, then I will focus on Lepton Flavour universality (LFU) violation in tau decays using multivariate analysis methods. LFU requires that the three charged lepton couple to the W bosons in the same way. This means that any ratio between the couplings ge, gμ and gτ , should equal the unity. By measuring those ratios, one can possibly reject the standard model of parti- cle physics. By performing high precision measurements, which where done by the the Belle II detector, located at the SuperKEKB accelerator in Tsukuba, Japan, one can determine the couplings. Multivariate analysis methods help in the Signal/Background classification which leads to a bigger phase space and ultimately to a more precise de- termination of the coupling ratios. The investigation of LFU violation is done in τ → lν ̄νdecays using 3x1 prong τ -pair events at Belle II. The algorithm is trained on MC14 data and the purity and efficiency metric will then be compared to a similar analysis which uses a cut based approach.
000004233 700__ $$aChristoph Schwanda$$edir.
000004233 700__ $$aGianluca Inguglia$$edir.
000004233 8560_ $$fchristoph.schwanda@oeaw.ac.at
000004233 8564_ $$uhttps://docs.belle2.org/record/4233/files/BELLE2-MTHESIS-2024-008.pdf
000004233 980__ $$aTHESIS