Graph neural networks for lepton identification in CMS experiment

Supervisor: Suman Chatterjee and Robert Schöfbeck

Leptons play a key role in many precision measurements and searches for physics beyond the Standard Model. Conventional methods for identifying leptons, electrons and muons at the CMS experiment at CERN LHC, however, use simple measures of hadronic activity for background suppression and, therefore, quickly reach their limits. Therefore, a new strategy is required to distinguish leptonic W, Z, or Higgs boson decays from the background.

For the further development of a new algorithm based on Graph Neural Networks (GNNs), we are looking for Master's students interested in gaining new experiences in the rapidly advancing field of machine learning.

The tasks are:

* Developing a GNN-based algorithm for identifying and distinguishing between muons and electrons of different origins.

* Calibration using collision data from the CMS experiment.

Experience in Python and/or C/C++, as well as basic knowledge of particle physics, is an advantage for this project.