Search for the Higgs boson decaying to bottom quarks and $W$-boson tagging techniques at the ATLAS experiment at the LHC

The Standard Model of particle physics is currently the most complete theory of subatomic particles. The discovery of the Higgs boson with a mass of 125 GeV in 2012 further validated the Standard Model, providing evidence for the theory that vector bosons obtain non-zero masses through the Higgs mechanism. Studies are ongoing to determine the exact nature and properties of the Higgs boson. A Higgs boson of this mass is predicted to decay to a pair of $b \bar b$ quarks with a branching ratio of 58%, however, this decay mode has not yet been observed. This thesis presents a search for the associated production of a Higgs boson with a leptonically decaying $W$-boson, $W H \rightarrow \ell \nu b \bar b$ using $20\;\mathrm{fb}^{−1}$ of Run 1 data collected by ATLAS at the LHC from pp collisions at a centre-of-mass energy of $\sqrt{s}=8\;\mathrm{TeV}$. The observed (expected) significance of a Higgs boson with a mass of 125 GeV for the $W H \rightarrow \ell \nu b \bar b$ process is found to be $2.7\sigma$ ($1.3\sigma$). The measured cross section in units of the expected Standard Model cross section has a best-fit value of $\mu=\frac{\sigma}{\sigma_{\mathrm{SM}}} = 2.2 ^{+0.67}_{−0.64}(stat.) ^{+0.7}_{−0.59}(syst.) = 2.2^{+0.97}_{−0.87}$. The results are combined with the search for $Z H \rightarrow \ell \ell b \bar b$ and $Z H \rightarrow \nu\bar{\nu} b \bar b$ to provide a best-fit value of $\mu=\frac{\sigma}{\sigma_{\mathrm{SM}}} = 1.1 ^{+0.61}_{−0.56}$. The start of Run 2 of the LHC in 2015 saw the collision energy being raised to $\sqrt{s}= 13\;\mathrm{TeV}$, increasing the probability of particles being produced with a large momentum boost. At these high energies there is also a possibility to discover new particles and interactions. An extension of the Standard Model, the Heavy Vector Triplet (HVT) model, describes new heavy vector bosons $W'$ and $Z'$, which can decay to pairs of heavy bosons ($W$, $Z$ or Higgs bosons). If the $W'$ and $Z'$ bosons are sufficiently heavy, the hadronic decays of the diboson final states produce boosted jets. In this thesis, methods for identifying hadronically decaying boosted bosons are developed, based on techniques that examine the internal substructure of the jet. Multiple substructure variables are combined into a single discriminant using two machine learning techniques: boosted decision trees and deep neural networks. Simulated events of $W'\rightarrow W Z \rightarrow qq \bar q \bar q$ are used to develop these "boosted $W$-boson taggers". An improvement in the background rejection power, whilst keeping 50% of the signal, over previous boosted $W$=boson taggers of up to 13% – when using deep neural networks – and 36% – when using boosted decision trees–is obtained. The performance of the new boosted $W$-boson taggers are evaluated in a search for a narrow WW resonances from the decay of a $Z'$ with boson-tagged jets in $3.2\;\mathrm{fb}^{−1}$ of $pp$ collisions at $\sqrt{s}=13\;\mathrm{TeV}$ collected with the ATLAS detector.