Title: Real time analysis strategies for reconstruction, exotic physics, and market analysis
Description: Real time hadronic reconstruction in LHC experiments is particularly difficult because of extremely busy detector images created by the multiple proton interactions (pile-up) in each bunch collision. This challenge will only increase in the future LHC upgrade. ESR5’s first objective is evaluate ML techniques for real-time hadronic reconstruction in ATLAS, as a replacement to algorithms that are too slow to be used in the trigger. The student in this project will be trained in reconstruction, modern ML techniques, and general classification tools. This expertise will be crucial for the second objective of this project within a collaboration with LIGHTBOX, in which the student will utilise ML techniques for predictive analytics on market analysis. The third objective of this project will be to evaluate GPUs for hadronic real-time reconstruction at the higher pile-up conditions of the LHC upgrades. The student will compare GPU-optimized reconstruction to CPU-based reconstruction. The student will receive dedicated co-supervision in optimizing algorithms for modern computing architectures and GPUs. The student will then apply their knowledge to specialised hadronic signatures for displaced or delayed jets in ATLAS, one of the most promising and experimentally challenging NP signatures. The student will search for exotic long-lived particle (LLP) signatures with this selection will be a significant step beyond ATLAS’s current capabilities, answer crucial questions for the future of ATLAS and open new avenues in the search for new physics.
Host country: Switzerland
Host beneficiary: University of Geneva
PhD-awarding institution: University of Geneva
Planned collaborations: LIGHTBOX, University of Santiago
