Search for Electroweak Supersymmetry in final states with a W boson and a Higgs boson with the ATLAS detector.

Vicentini, Silvia (2025) Search for Electroweak Supersymmetry in final states with a W boson and a Higgs boson with the ATLAS detector. [Laurea], Università di Bologna, Corso di Studio in Fisica [L-DM270]
Documenti full-text disponibili:
[thumbnail of Thesis] Documento PDF (Thesis)
Disponibile con Licenza: Creative Commons: Attribuzione - Non commerciale - Non opere derivate 4.0 (CC BY-NC-ND 4.0)

Download (3MB)

Abstract

Supersymmetry is one of the most promising extensions of the Standard Model, addressing several of its theoretical limitations and providing a natural dark matter candidate in the form of the lightest supersymmetric particle. Among the various Supersymmetry production mechanisms, the electroweak production of charginos and neutralinos is of particular interest due to its cleaner experimental signatures. This work focuses on the simplified Supersymmetry process, analyzing final states with one or two charged leptons, using Monte Carlo simulations of events reconstructed by the ATLAS detector. The goal of the analysis is to discriminate Supersymmetry signal from the SM backgrounds in different final states, taking into account the Standard Model branching ratios of the $W$ and Higgs bosons. Signal regions are defined and optimized by studying the statistical significance in the single-lepton and dilepton channels. The dilepton channel is explored for the first time in this context. Although the expected significance is lower than the single-lepton channel, due to the reduced branching ratio, newly defined signal regions allow for an improvement in sensitivity. In the single-lepton channel, several analysis strategies were employed. A Variational Autoencoder was used to detect signal-like anomalies, demonstrating good performance in scenarios with large mass splitting. In addition, both Cut and Count and Deep Neural Network techniques were applied and compared. The strategy based on the Deep Neural Network showed the best performance, yielding improved sensitivity for benchmark mass hypotheses. For instance, statistical significances of Z = 2.48 and Z = 1.79 were obtained for the (500,250) GeV and (800,0) GeV mass hypotheses, respectively. The results highlight the potential of machine learning methods, particularly Deep Neural Networks, to enhance sensitivity in searches for supersymmetric particles produced through the electroweak interaction.

Abstract
Tipologia del documento
Tesi di laurea (Laurea)
Autore della tesi
Vicentini, Silvia
Relatore della tesi
Scuola
Corso di studio
Ordinamento Cds
DM270
Parole chiave
Supersymmetry,charginos,neutralinos,Cut and Count,DNN,VAE,ATLAS,LHC
Data di discussione della Tesi
19 Settembre 2025
URI

Altri metadati

Statistica sui download

Gestione del documento: Visualizza il documento

^