Optimization of a Run 3 Search for Higgs Decays to Dark Photons using the ATLAS Detector at the Large Hadron Collider

Despite the remarkable success of the Standard Model (SM), it has yet to incorporate any candidates for dark matter, which we know exists from astrophysical observations. To search beyond the Standard Model, we postulate that dark matter may be part of a broader “dark sector”. By introducing messenger fields charged under both SM and this dark sector, allowing the potential existence of massless dark photons. This analysis is conducted using proton-proton collision data using the ATLAS Detector at the Large Hadron Collider. Since gluon-gluon fusion is the most dominant Higgs production mode, we focus on this channel to search for the Higgs boson decaying to a photon and a dark photon. The experimental signature is SM photon, missing transverse energy (from the dark photon that does not interact in our detector), and collimated sprays of hadrons called jets. My research focuses on optimizing the sensitivity of this signal event using an integrated luminosity of 135 fb-1 from ATLAS Run 3 data. A Machine Learning classifier study, including Boosted Decision Tree and Deep Neural Network approaches, was developed to enhance the signal significance. After optimizing the selection criteria, the analysis improved the expected upper limit on the branching ratio of this process compared to the previous work by 17% (from 0.92 to 0.76). This tighter upper limit rules out a larger portion of the theoretical parameter space than previous studies.