Van Stroud, Samuel John;
(2023)
Graph Neural Network Flavour Tagging and Boosted Higgs Measurements at the LHC.
Doctoral thesis (Ph.D), UCL (University College London).
Preview |
Text
svs_thesis_corrected.pdf - Accepted Version Download (9MB) | Preview |
Abstract
This thesis presents investigations into the challenges of, and potential improvements to, b-jet identification (b-tagging) at the ATLAS experiment at the Large Hadron Collider (LHC). The presence of b-jets is a key signature of many interesting physics processes such as the production of Higgs bosons, which preferentially decay to a pair of b-quarks. In this thesis, a particular focus is placed on the high transverse momentum regime, which is a critical region in which to study the Higgs boson and the wider Standard Model, but also a region within which b-tagging becomes increasingly difficult. As b-tagging relies on the accurate reconstruction of charged particle trajectories (tracks), the tracking performance is investigated and potential improvements are assessed. Track reconstruction becomes increasingly difficult at high transverse momentum due to the in- creased multiplicity and collimation of tracks, and also due to the presence of displaced tracks from the decay of a long-flying b-hadron. The investigations reveal that the quality selections applied during track reconstruction are suboptimal for b-hadron decay tracks inside high transverse momentum b-jets, motivating future studies into the optimisation of these selections. Two novel approaches are developed to improve b-tagging performance. Firstly, an algorithm which is able to classify the origin of tracks is used to select a more optimal set of tracks for input to the b-tagging algorithms. Secondly, a graph neural network (GNN) jet flavour tagging algorithm has been developed. This algorithm directly accepts jets and tracks as inputs, making a break from previous algorithms which relied on the outputs of intermediate taggers. The model is trained to simultaneously predict the jet flavour, track origins, and the spatial track-pair compatibility, and demonstrates marked improvements in b-tagging performance both at low and high transverse momenta. The closely related task of c-jet identification also benefits from this approach. Analysis of high transverse momentum H → bb decays, where the Higgs boson is produced in association with a vector boson, was performed using 139 fb−1 of 13 TeV proton-proton collision data from Run 2 of the LHC. This analysis provided first measurements of the V H, H → bb process in two high transverse momentum regions, and is described with a particular focus on the background modelling studies performed by the author.
| Type: | Thesis (Doctoral) |
|---|---|
| Qualification: | Ph.D |
| Title: | Graph Neural Network Flavour Tagging and Boosted Higgs Measurements at the LHC |
| Open access status: | An open access version is available from UCL Discovery |
| Language: | English |
| Additional information: | Copyright © The Author 2023. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request. |
| UCL classification: | UCL UCL > Provost and Vice Provost Offices > UCL BEAMS UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Maths and Physical Sciences UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Maths and Physical Sciences > Dept of Physics and Astronomy |
| URI: | https://discovery-pp.ucl.ac.uk/id/eprint/10173308 |
Archive Staff Only
 |
View Item |
