A new paper led by Yilun Wu, published in the Journal of High Energy Physics, brings together her work on using machine learning to understand how energetic particles lose energy in hot nuclear matter.
When heavy atomic nuclei collide, they briefly create quark–gluon plasma—a state of matter in which quarks and gluons move beyond the confines of individual protons and neutrons. Sprays of particles called jets can lose energy as they travel through this material, a process known as jet quenching.
The team trained a neural network to recognize patterns in a jet’s internal structure. In simulated collisions, its predictions closely tracked how much energy individual jets lost. The method remained effective when simulated detector effects were included, and further checks used quantities the network had not been trained on.
These results support using machine learning to identify quenched jets and study their interactions with the plasma.
The publication marks an important milestone in Yilun’s research, bringing together her efforts to develop and carefully validate this approach. Congratulations, Yilun!