Looking for phenomena beyond the Standard Model with the ATLAS experiment at the LHC

As convener of the Exotics group at the very beginning of the LHC operations, I contributed to developing a comprehensive program of searches for phenomena beyond the Standard Model (BSM). In Chapter 2, I give a brief overview of this vast program and review the most important results. In ATLAS, Supersymmetry is studied in a group distinct from the Exotics group; for completeness I also describe succinctly the strategy and the recent results of the search for Supersymmetry. Given the great diversity of BSM models and the experimental signatures they lead to, a methodical strategy had to be put in place, with the priority given to the observation of any deviation from the Standard Model, independently of the underlying cause for such a deviation: the emphasis was put on analyses that cover a wide range of models rather than being optimized for a specific one. As much as possible, all experimental signatures were explored, independently of their theoretical motivation. Chapter 3 describes a specific BSM search to which I contributed directly and is based on an ATLAS article that was published in 2015: the search for vector-like quarks (VLQ) produced in pair and involving a Higgs boson in their decay. I focused on the case of a pair of bottom-like heavy quarks decaying to bHbH that had been neglected in previous analyses. In Chapter 4, I describe two personal contributions to the identification of b-jets: soft-muon b-tagging, and the calibration of the b-tagging algorithms performance using top-antitop events. Soft-muon b-tagging makes use of the presence of a relatively large-momentum muon coming from the semi-leptonic decay of B hadrons in b-jets. Joining my expertise in b-tagging acquired during my PhD and the expertise of the Saclay group in muon reconstruction, I developed the first soft-muon tagger in ATLAS and commissioned it with the very first data accumulated in 2009. In parallel I worked on the first studies of techniques of calibration of the b-tagging performance using data. The large top pair production cross-section at the LHC permits to accumulate a large top pair control sample. Thanks to the well-known heavy flavor content in those events (two b-jets in each event, at tree level), the b-tagging efficiency can be measured in data with great precision and over a large range of jet momentum. I developed a technique that was able to give first results with very low integrated luminosity. The same technique is applied today, with large integrated luminosity, to measure the top pair production cross-section with in situ calibration of the b-tagging efficiency, thus removing the dominant systematic uncertainty from b-tagging and leading to the most precise measurement of this quantity. This work was described in two ATLAS public documents. Finally in Chapter 5, I discuss the future prospects of BSM searches at the LHC as well as my personal research projects. (author)