Study Towards Analysing the Higgs Self-Coupling in the $hh$ → 4τ Channel at the ATLAS Detector

The Higgs boson discovery at the LHC in 2012 completes the predicted particle content of the Standard Model (SM), leaving not much space for unknown physics in the accessibly TeV scale. This heralds the precision measurement era of SM parameters of which the Higgs self-coupling is one of the last unmeasured ones. Its high sensitivity to new physics scenarios further increases its relevance. Due to the very small cross sections, the power of the LHC to measure the self-coupling is estimated to be limited, which motivates the investigation of this process in as many channels as possible. Apart of more intensively studied decay channels, the $hh\rightarrow4\tau$ channel shapes up as a very promising additional decay, because a very limited amount of other processes is expected to produce four truth taus. However, a large background containing several sources of QCD jets can fake hadronically decaying $\tau$~leptons and has to be fought. This work studies the conspicuity of di-Higgs production within multi-tau events. As the signal efficiency is expected to be low due to the high amount of demanded $\tau$~leptons, a novel reconstruction and identification method is developed. Instead of reconstructing every $\tau$~lepton candidate on its own, the whole event topology is studied to separate processes with many truth taus from fake enriched ones. It is shown that the new method enables either a doubling of the signal significance or of the signal efficiency, while the corresponding other value remains at its value when using classical reconstruction methods. All relevant background processes are included in this thesis. To that effort, a sophisticated data-driven fake estimation is developed for four objects. This method is aligned to the used tau combination method, although both approaches are based on the lepton's charges. Finally, three different cut-based signal regions are designed by demanding either none, one or two light leptons together with the corresponding amount of hadronically decaying taus (\fhzl, \thol and \thtl). The signal selection is further improved by exploiting boosted decision trees. Overall, a combined significance of $\sigma = 0.89$ for an integrated luminosity of \SI{3}{\per\atto\barn} is achieved. This value is comparable to the expected significance of the widely studied $hh\rightarrow 4b$ channel of $\sigma = 0.6$, although for the $hh\rightarrow b\bar{b}\tau\tau$ ($\sigma = 2.1$) and $hh\rightarrow b\bar{b}\gamma\gamma$ ($\sigma = 2.0$) channels even higher values are expected. Eventually, an upper limit of the production cross section of $\sigma_\text{ggF}/\sigma_\text{ggF}^\text{SM} = 2.40^{+0.95}_{-0.61}$ is predicted for a dataset of \SI{3}{\per\atto\barn} at \SI{95}{\percent} confidence level. To compare this result to published channels, the limit is recalculated to be $\sigma_\text{ggF}/\sigma_\text{ggF}^\text{SM} = 36.9^{+22.5}_{-13.5}$ for \SI{36.1}{\per\fb}. This value outperforms the upper limits of published decay channels containing $W$ bosons ($\sigma_\text{ggF}/\sigma_\text{ggF}^\text{SM} = 120 - 300$) by far. Instead, it is more comparable to the exclusion capability of the previously named most promising decay channels ($\sigma_\text{ggF}/\sigma_\text{ggF}^\text{SM} = 12 - 25$). Hence, it is expected that the studied $hh\rightarrow 4\tau$ channel is capable to valuably contribute to the overall measurement of the Higgs self-coupling at the ATLAS detector and it should be investigated further in future.