Speaker
Description
The search for dark matter motivates searches for new particles beyond the Standard Model (SM), such as the dark photon $\gamma_d$. The work described in this presentation focuses on the search for massless $\gamma_d$ via a hypothetical Higgs boson decay $H\rightarrow Z+\gamma_d$, with $Z\rightarrow l^+l^-$. The non-interacting nature of the dark photon would give arise to a missing transverse momentum ($E_T^{miss}$), subsequently leading to a $l^+l^-+\ E_T^{miss}$ final state in the detector. This study will present the first search of such process at the LHC using 140 ${fb}^{-1}$ of proton-proton collisions collected with the ATLAS detector at 13.6TeV centre-of-mass energy. Several Machine Learning (ML) algorithms were studied to enhance the sensitivity of the search and optimize the dark photon signal acceptance and SM background processes rejection. Monte Carlo Simulated Data were used to efficient classification and define a signal region (SR) where a potential excess of events with respect to SM predictions could be observed in data. Both Boosted Decision Tree (BDT) and Recurrent Neural Network (RNN) were used for an optimal SR selection. In addiction Control regions for dominants backgrounds from SM processes (Z+jets, $t\bar{t}$,$ll\nu\nu$) were defined for background estimation validation and data/MC comparison. A first estimation of the sensitivity to the signal from $H\rightarrow Z+\gamma_d$ as well as a limit on the branching ratio BR($H\rightarrow Z+\gamma_d$) are derived.
| Apply for student award at which level: | MSc |
|---|