Thesis

Anomaly detection in the search for new physics with the ATLAS detector

• Call:

  PT-CERN Call 2022/1

• Academic Year:

  2022

• Domain:

  Astroparticle Physics

• Supervisor:

  Inês Ochoa

• Co-Supervisor:

  Patricia Conde Muino

• Institution:

  Instituto Superior Técnico (Universidade de Lisboa)

• Host Institution:

  Laboratório de Instrumentação e Física Experimental de Partículas

• Abstract:

  This project foresees the application of anomaly detection methods in the search for new physics in proton-proton collisions with the ATLAS detector at the Large Hadron Collider. ATLAS detector data in final states with high-energy jets will be analysed. The candidate will explore the new Run 3 dataset, to be collected starting in 2022, looking for signs of new resonances by taking advantage of state-of-the-art machine learning techniques for anomaly detection that can identify rare signals and distinguish them from background processes with cross-sections that are several orders of magnitude larger. Unsupervised and weakly supervised searches will be examined, with the aim of ensuring a robust and model-independent exploration of the dataset. The study and implementation of novel machine-learning methods for data-driven background estimation is also expected. The candidate will work with a team at LIP which has vast expertise in hadronic final states and that has led recent ATLAS publications in searches for new physics with these final states. The candidate is also expected to contribute to the successful operation of the experiment, in particular in the trigger and jet performance. The candidate is expected to develop their research alongside members of other institutions that are part of the ATLAS Collaboration and to travel to CERN to present their work.