Deutsch Intern
  • [Translate to Englisch:] Seminar Elementarteilchenphysik
Theoretical Physics II

Dr. Miguel Crispim Romão

Durham University

T a l k : 16. May 2024

Machine Learning and the Quest to Isolate Jets Quenched by the Quark Gluon Plasma

Abstract

In this seminar I will present advancements in the study of Quark Gluon Plasma (QGP) through the examination of jet modifications in ultrarelativistic collisions of heavy ions using modern Machine Learning (ML). In the first study, Deep Learning techniques were employed to identify jets strongly influenced by QGP interactions. Utilising simulated Z+jet events in vacuum (pp collisions) and medium (PbPb collisions), dedicated Convolutional Neural Networks, Dense Neural Networks, and Recurrent Neural Networks were developed and trained. The results demonstrate the potential of these techniques for discriminating between medium- and vacuum-like jets within the PbPb sample, revealing insights into jet quenching effects induced by the QGP. In the second study, a comprehensive set of jet substructure observables was surveyed to analyse modifications resulting from QGP interactions. Using pp and PbPb simulated dijet events, ML techniques were applied to distinguish vacumm- and medium-like jet samples. The analyses revealed high correlations between observables, with specific pairs capturing the full sensitivity to quenching effects, contributing to a deeper understanding of QGP-related phenomena in heavy-ion collisions. Finally, I will present ongoing work on producing a state-of-the-art discriminant between vacumm- and medium-like jets, answering the question of whether it is possible to completely isolate jets quenched by the QGP.