Measurement of track-based missing transverse momentum in proton-proton collisions at $\sqrt{s}=8$~TeV centre-of-mass energy with the ATLAS detector

The missing transverse momentum ($\mathbf{E}^{\mathrm{miss}}_{\mathrm{T}}$) in experiments at particle colliders such as the Large Hadron Collider (LHC) at CERN is defined as the momentum imbalance in the plane transverse to the beam axis: the resultant of the negative vectorial sum of the momenta of all the particles that are involved in the $pp$ collision of interest. The reconstruction of the $\mathbf{E}^{\mathrm{miss}}_{\mathrm{T}}$ to the high precision necessary for physics analyses involves combining contributions from the fully reconstructed and calibrated hard objects in the event, most of which rely on energy measurements in the calorimeters to provide some or all of the information about the object. Additionally, contributions from particles that do not pass the hard object $\mathbf{p}_{\mathrm{T}}$ thresholds form a "soft" term which is traditionally built using the topological clusters from unmatched energy deposits in the calorimeter, a method which has historically served collider experiments well, but one which is susceptible to fluctuations from noise and particles from additional, unrelated collisions within the same event. In this thesis we document the development of a new approach to the missing transverse momentum reconstruction within the ATLAS experiment, using only charged particle tracks reconstructed by the ATLAS Inner Detector to provide us with a tracks-only measurement of the missing transverse momentum which we denote $\mathbf{p}^{\mathrm{miss}}_{\mathrm{T}}$. An optimised track selection procedure was developed and used to select only high quality tracks originating from the signal $pp$ collision vertex to form the equivalent soft term of the $\mathbf{p}^{\mathrm{miss}}_{\mathrm{T}}$. For hard objects such as electrons, muons, and jets, two methods are investigated: one using the tracks-only approach, and the other using the reconstructed, calibrated objects as used by the $\mathbf{E}^{\mathrm{miss}}_{\mathrm{T}}$. The combination of the tracks-only based soft term and the hard objects together form what we call the nominal and object-corrected versions of the $\mathbf{p}^{\mathrm{miss}}_{\mathrm{T}}$ respectively. Detailed studies of the $\mathbf{p}^{\mathrm{miss}}_{\mathrm{T}}$ algorithms were performed on 8 TeV $pp$ collision data collected during the LHC 2012 run and are presented here. These studies, which cover the performance of the $\mathbf{p}^{\mathrm{miss}}_{\mathrm{T}}$ in various event configurations and comparisons with Monte Carlo predictions, demonstrate a robustness of the $\mathbf{p}^{\mathrm{miss}}_{\mathrm{T}}$ against the same noise and pileup fluctuations that threaten the $\mathbf{E}^{\mathrm{miss}}_{\mathrm{T}}$; a result that has already garnered the use of the $\mathbf{p}^{\mathrm{miss}}_{\mathrm{T}}$ in a number of ATLAS physics analyses.