Open Collections

Chisholm, Jenna Lori

University of British Columbia

The Standard Model (SM) is a theory that describes the fundamental particles of our universe and their interactions. However, despite its incredible success, there is still an array of phenomena that the SM fails to explain, instigating the search for new physics that could confirm theories Beyond the Standard Model (BSM). One particle that often plays a special role in many of these BSM theories is the heaviest known fundamental particle: the top quark. Reconstruction of top quarks produced in high energy particle collisions — such as those of the Large Hadron Collider (LHC) — to the best possible resolution is therefore crucial; from improved mass resolutions for bump hunting to more diagonal unfolding matrices for differential cross-section measurements, such improvements will enhance our sensitivity to BSM effects in both precision measurements and searches for new physics. This thesis presents a newly designed deep neural network (TRecNet) for the ATLAS experiment that infers the four-vectors of the top and anti-top quarks from detector-level decay products in the semi-leptonic decay channel of top anti-top pair production. The performance of TRecNet and several slight variations of the network are compared to traditional top reconstruction algorithms that are based on kinematic constraints and likelihood fits. The neural networks are shown to consistently improve upon the reconstruction precision in comparison to the likelihood-based methods, in addition to obtaining this improved precision more efficiently.

Text

2023-10-19

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/86232

Physics

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-nd/4.0/