Into the hidden valley: probing experimental signatures of strongly interacting dark matter models
Date
2023
Authors
Wilson, Danielle Joan
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The lack of preeminent theory guiding particle physics, along with the increasingly large amounts of data being produced at the Large Hadron Collider, requires resources to be
innovatively utilised in the solving of the field’s outstanding problems. This work is an exploration into strongly coupled dark sector models - which exhibit parton showering akin to that of QCD. Constraints on different parametrisations of semivisible jet s- and t- channel final states were obtained using the Contur [1] tool - which employs the hundreds of existing measurements stored in Rivet [2] to assess the viability of new theories. Upper limits on the semivisible t-channel mediator mass were obtained, using Contur, that were of the same order of magnitude as the upper limits obtained in the dedicated search performed Ref. [3]. In order to illuminate semivisible jets against the vast background, the efficacy of using Convolutional Neural Networks to discriminate semivisible jet images was explored. This was the first study of this nature to use ATLAS detector-level semivisible jet t-channel samples. The results - although preliminary in nature, due to the relatively small sample sizes - displayed excellent classification performance according to the AUC scores.
Description
A dissertation submitted in fulfillment of the requirements for the degree of Master of Science to the Faculty of Science, School of Physics, University of the Witwatersrand, Johannesburg, 2022