Events in our system are self-managed.  Group and event managers are encouraged to review privacy and security settings, and adjust them if needed.  If you need assistance please contact Indico Support - contact Help at bottom of page. https://learn.getindico.io/categories/managing/

15–18 Feb 2024
Mt. Bromont
Canada/Eastern timezone

Identifying New Long-Lived Particles (LLPs) Using Graph Neural Networks with the ATLAS Detector

17 Feb 2024, 09:15
15m
Salon AB (Hotel Chateau Bromont)

Salon AB

Hotel Chateau Bromont

Contributed Oral Morning 3 - Feb. 17, 2024

Speaker

Paras Pokharel (Simon Fraser University)

Description

Many extensions to the Standard Model (SM) predict the existence of strongly interacting dark sectors, which behave similarly to Quantum Chromodynamics (QCD) but interact weakly with SM particles. Depending on the parameters of the dark sector, one potential LLP signature could be that of an "emerging jet" – a spray of particle tracks after proton-proton collision dominantly composed of displaced tracks and containing many displaced vertices within the jet cone. Graph Neural Networks (GNN) have shown great promise in capturing complex dependencies and patterns in graph-structured data, making them well-suited for analyzing the intricate topology of emerging jets. GNN-based architecture is used to tag emerging jets with high accuracy while significantly suppressing the QCD background. The architecture of the GNN also enables the classification of displaced tracks as well as the identification of displaced vertices within the jet cone. I will present a comprehensive evaluation of the GNN's performance in these jet, track, and vertex-level classification tasks providing valuable insight into the long-lived particle decay signatures.

Your current academic level, MSc student
Your Email ppa45@sfu.ca
Affiliation Simon Fraser University
Supervisor Matthias Danninger
Supervisor Email mdanning@sfu.ca

Primary author

Paras Pokharel (Simon Fraser University)

Presentation materials