Jet Calibration: A Regression and Classification Model for the ATLAS Global Event Processor

Measurement of the Higgs self-coupling is one of the core objectives of the ATLAS experiment for the CERN laboratory’s High-Luminosity Large Hadron Collider (2029-2045). The di-Higgs process is sensitive to the jet momentum threshold available from the trigger system. One possible solution for lowering the threshold is to implement machine learning algorithms in the ATLAS Global Event Processor, part of the FPGA hardware trigger. These small model architectures allow for accurate and resource-efficient jet reconstruction on the scale of nanoseconds when run on FPGA hardware. In particular, this project uses a regression and classification boosted decision tree model to implement a jet calibration scheme using simulated events of the ATLAS detector.