Searching for Axion-Like Long-Lived Particles at the ATLAS Experiment and a Contribution to the MadAnalysis5 Recasting Framework

If general relativity is a valid description of gravity, Dark Matter (DM) makes up about 27% of the universe's energy content, posing one of the most profound ques- tions in modern physics. Numerous Beyond Standard Model (BSM) theories propose new particle candidates and mediators for dark matter, including weakly interact- ing particles (WIMPs) and axion-like particles (ALPs) as prominent examples that have possible masses spanning over 30 orders of magnitude. ALPs, usually described in the framework of effective field theories, could exhibit long lifetimes, leading to long-lived particles (LLPs) possibly acting as mediators to the dark sector. These LLPs could generate exotic signals in collider experiments like ATLAS and CMS, challenging conventional identification and reconstruction algorithms. This thesis focuses on developing novel calorimeter-based techniques to detect ALPs decaying into photon pairs within the ATLAS detector, addressing the unique challenges posed by displaced decay vertices to enhance the sensitivity of LLP searches. In addition to the primary focus on dark matter and LLP searches, this work in- cludes significant contributions to the MadAnalysis 5 (MA5) framework for rein- terpreting ATLAS and CMS experimental results. The framework now supports efficient ML-based analysis recasting by integrating machine learning (ML) models through the ONNX inference software utility. This development facilitates the ac- curate reproduction and reinterpretation of complex analyses that use ML-driven event selection methods and ensures consistent performance across different envi- ronments. As a demonstration of this enhanced capability, an ATLAS search for R-parity violating Supersymmetry has been reproduced within MA5. This project exhibits possible progress in the way the LHC experiments publish auxiliary ma- terial used to allow easy reinterpretation and provide interesting legacy data for future LHC analyses recasting. Furthermore, this thesis discusses the upgrades to ATLAS subsystems to prepare for the HL-LHC run and details the progress in pixel detector technology accomplished through the future Inner Tracker (ITk) test-beam campaigns conducted in 2022. These campaigns evaluated sensor performances, achieving detection efficiencies ex- ceeding 98%. The insights gained were crucial in establishing better reconstruction software and workflows to guarantee consistency and robustness in the measured performances. Overall, this thesis presents a comprehensive approach to advancing ALP research through new LLP detection techniques, enhanced recasting capabilities for ML- based analyses, and contributions to detector technology development.