Fast particle tracking in the HL-LHC trigger is a crucial element of the success of the ATLAS physics program and in particular of RTA searches for dark matter, since one of the main backgrounds to these analyses is due to simultaneous proton-proton interactions that can be distinguished using tracking information. For this reason, this project consists of the implementation of the experiment-independent ACTS tracking software on accelerators, supervised by ATLAS experts at the University of Manchester and at CERN.
A second objective of this project is to implement anomaly detection on accelerators for the ATLAS experiment. Anomaly detection is an ML technique concerned with the detection of ultra-rare “anomalous” events which do not follow part of the “normal” pattern of input samples and where little is known about the distribution of these anomalies. In particular, there has been a growing interest in running those algorithms on accelerators such as FPGAs, to be able to use them already at the initial level of the trigger system in RTA. The student will be working with CERN experts to contribute to and use the HLS4ML software package for this purpose. The algorithms that are the outcome of this collaboration will then be used to refine the initial selection of the prospective dark matter search that uses accelerated tracking in subsequent trigger levels.
The student working on this project will also be integrated in the interdisciplinary 4IR centre for doctoral training at the University’s Institute for Data Science and AI led by Manchester scientists.
Host country: United Kingdom
Host beneficiary: University of Manchester
PhD-awarding institution: University of Manchester
Planned collaborations: CERN