%0 Generic
%A Sauer, Christof
%C Heidelberg
%D 2023
%F heidok:33941
%R 10.11588/heidok.00033941
%T Measurement of the Triple-Differential Cross-Section for the Production of Multijet Events using 139 fb^{-1} of Proton-Proton Collision Data at \sqrt{s} = 13 TeV with the ATLAS Detector to Disentangle Quarks and Gluons at the Large Hadron Collider
%U https://archiv.ub.uni-heidelberg.de/volltextserver/33941/
%X At hadron-hadron colliders, it is almost impossible to obtain pure samples in either quark-  or gluon-initialized hadronic showers as one always deals with a mixture of particle jets.  The analysis presented in this dissertation aims to break the aforementioned degeneracy by  extracting the underlying fractions of (light) quarks and gluons through a measurement of the  relative production rates of multijet events.    A measurement of the triple-differential multijet cross section at a centre-of-mass energy of  13 TeV using an integrated luminosity of 139 fb −1 of data collected with the ATLAS detector  in proton-proton collisions at the Large Hadron Collider (LHC) is presented. The cross section  is measured as a function of the transverse momentum p T , two categories of pseudorapidity  η rel defined by the relative orientation between the jets, as well as a Jet Sub-Structure (JSS)  observable O JSS , sensitive to the quark- or gluon-like nature of the hadronic shower of the two  leading-p T jets with 250 GeV < p T < 4.5 TeV and |η| < 2.1 in the event.  The JSS variables, which have been studied within the context of this thesis, can broadly be  divided into two categories: one set of JSS observables is constructed by iteratively declustering  and counting the jet’s charged constituents; the second set is based on the output predicted by  Deep Neural Networks (DNNs) derived from the “deep sets” paradigm to implement permutation  invariant functions over sets, which are trained to discriminate between quark- and gluon-  initialized showers in a supervised fashion.    All JSS observables are measured based on Inner Detector tracks with p T > 500 MeV  and |η| < 2.5 to maintain strong correlations between detector- and particle-level objects.  The reconstructed spectra are fully corrected for acceptance and detector effects, and the  unfolded cross section is compared to various state-of-the-art parton shower Monte Carlo  models. Several sources of systematic and statistical uncertainties are taken into account that  are fully propagated through the entire unfolding procedure onto the final cross section. The  total uncertainty on the cross section varies between 5 % and 20 % depending on the region of  phase space.    The unfolded multi-differential cross sections are used to extract the underlying fractions  and probability distributions of quark- and gluon-initialized jets in a solely data-driven, model-  independent manner using a statistical demixing procedure (“jet topics”), which has originally  been developed as a tool for extracting emergent themes in an extensive corpus of text-based  documents. The obtained fractions are model-independent and are based on an operational  definition of quark and gluon jets that does not seek to assign a binary label on a jet-to-jet basis,  but rather identifies quark- and gluon-related features on the level of individual distributions,  avoiding common theoretical and conceptional pitfalls regarding the definition of quark and  gluon jets.    The total fraction of gluon-initialized jets in the multijet sample is (IRC-safely) measured  to be 60.5 ± 0.4(Stat) ⊕ 2.4(Syst) % and 52.3 ± 0.4(Stat) ⊕ 2.6(Syst) % in central and forward  region, respectively. Furthermore, the gluon fractions are extracted in several exclusive regions  of transverse momentum.