URAP

Allison Xu

Graph Neural Network-based Top Quark Reconstruction

The Large Hadron Collider collects a large amount of data through collisions between photons, which produce and decay into various particles such as top quarks. Top quarks are the elementary particle with the most mass and could help in the search for new physics. Due to their mass, they decay quickly and are not detected directly. Instead, top quarks can be reconstructed using other observable information with machine learning. Graph neural networks (GNN) are useful for top quark reconstruction because of its ability to represent particles and their relationships as graphs. I will continue to develop GNNs, understand its performance, and explore the flexibility it provides in the representation of data.

Message to Sponsor

The URAP program and sponsors have provided me the incredible opportunity to contribute to an important area of research and to further my education in meaningful ways. This summer I will continue to gain more experience in machine learning for physics. I am thankful to URAP and the sponsors for supporting our research over the summer.
  • Major: Computer Science & Statistics
  • Mentor: Haichen Wang, Physics