Search for new phenomena in dijet events with quark/gluon tagger using the ATLAS detector

The Standard Model (SM) of particle physics is a successful theory that provides a comprehensive framework for understanding the fundamental forces and particles in the universe. It describes the electromagnetic, weak, and strong nuclear interactions, and includes a range of elementary particles like quarks, leptons, and the Higgs boson. Despite its huge success in explaining many aspects of particle physics, the SM has its limitations. There are various phenomena, such as dark matter and neutrino masses, that remain unexplained within the SM. These limitations motivates a quest for new physics beyond the Standard Model (BSM). Quantum Chromodynamics (QCD) is a pivotal component of the SM, representing the theory that describes the strong interaction. This interaction is responsible for bind- ing quarks and gluons, the basic constituents of protons, neutrons, and other hadrons. In high-energy collisions, such as those in the Large Hadron Collider (LHC), QCD dy- namics play a critical role. When quarks and gluons are scattered by collisions, they hadronised as jet and recorded by the detector. Quark and gluon jets are central to understanding many processes within particle physics. However, distinguishing between a jet originating from a quark and one from a gluon presents a significant challenge due to the hadronisation. The physics analyses presented in this PhD dissertation search for new BSM physics in the dijet mass dis- tribution using an integrated luminosity of 140 fb−1 of proton-proton collisions with a centre-of-mass energy at 13 TeV recorded by the ATLAS detector from the LHC. Many BSM models of new physics predict the particles could decay into a pair of gluons, this analysis aims at searching for the evidence of the BSM resonances that decay into two jets, with a specialized gluon-tagging approach adopted. This method hinges on the analysis of the charged tracks associated with a jet, to enhance the ability to detect new resonances . The process of distinguishing jets arising from quarks and gluons, commonly known as quark/gluon tagging, is pivotal in this analysis. Thus the performance and the cal- ibration of two quark/gluon jet taggers is studied: one tagger based on the number of charged tracks associated with the jets (Ntrk), while the other employs a boosted decision tree (BDT) to combine various jet substructure observables. The BDT-tagger outper formed the Ntrk-tagger in distinguishing quark-jets from gluon-jets up to 1200 GeV pT, with both taggers showing comparable performance above this range. Differences be- tween data and Monte Carlo simulation of quark-jet tagging efficiency are provided for jet transverse momenta ranging from 500 GeV to 2 TeV, encompassing various tagger working points, range between 0.92 and 1.02. These come with a combined system- atic uncertainty of about 20%, predominantly caused by theoretical modelling uncer- tainty. A matrix method is performed on data to retrieve the quark/gluon fraction from quark/gluon-enriched subsamples, defined by the pseudorapidity of the jet. In the analysis, the QCD background is estimated by the statistical framework which parametrised the dijet mass spectrum. Three different spectra are defined: the untagged dijet invariant mass, one gluon-tagged (1-g tagged) and two gluon-tagged (2-g tagged) channels for BSM signal Graviton and Quantum Black Hole (QBH) models. Because no significant deviation is observed, an upper limits for the cross-section then are set at the 95% confidence level (CL) for the three categories on signal models of BSM physics. The observed lower limits for Graviton mass were 3.81 TeV (untagged), 4.01 TeV (1-g tagged), and 4.26 TeV (2-g tagged). For the QBH model, the limits were 9.88 TeV (untagged), 9.89 TeV (1-g tagged), and above 10.00 TeV (2-g tagged) at 95% CL. Gluon tagging, enhanced by 5% per tagged gluon, especially at high pT. By incorporat- ing gluon-tagging on event selections, the search limits are improved for BSM signals considered.