A search for tWZ production in the trilepton channel using Run 2 data from the ATLAS experiment

Master Thesis

2021

Permanent link to this Item
Authors
Supervisors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher
License
Series
Abstract
This dissertation describes an analysis of events containing three leptons from 136 fb−1 of proton-proton collision data, with a centre of mass energy of 13 TeV, recorded by the ATLAS detector between 2016 and 2018. The aim of the analysis was to search for evidence of a top quark produced in association with a W boson and a Z boson (tWZ). An event selection scheme was developed using simulation to broadly suppress background events and to select signal events. Events were separated into mutually-exclusive regions of phase space to increase the ratio of signal to backgrounds and to calibrate the modelling of the backgrounds. Background events were further suppressed through the use of the Gradient Boosted Decision Tree (GBDT) machine learning algorithm. First, a hadronically-decaying W boson candidate was identified using a GBDT; this was used to suppress WZ background events. Then, an event-level GBDT was used to suppress all background events. A maximum likelihood fit was used to estimate the signal strength µ of tWZ production, where nuisance parameters were assigned to theoretical and experimental systematic uncertainties. The work presented here forms the basis of an official ATLAS experiment analysis, thus, the signal region was blinded to avoid potential biases in the future development of the official ATLAS analysis. The best fit value of µ resulting from a fit to a modified Asimov dataset was µˆ = 2.08+1.48 −1.45. This corresponds to an expected significance of Z exp µ = 0.72 σ. An expected upper limit µ exp up = 2.77+2.39 −1.28 was also determined from the fit to the modified Asimov dataset. Thus, this analysis has the potential to put the strongest ever constraint on tWZ production, but does not have the potential to observe tWZ production as predicted in the Standard Model. These constraints are limited by statistical uncertainties.
Description
Keywords

Reference:

Collections