Deep Learning for the classification of quenched jets

Abstract

An important aspect of the study of Quark-Gluon Plasma (QGP) in ultrarelativistic collisions of heavy ions is the ability to identify, in experimental data, a subset of the jets that were strongly modified by the interaction with the QGP. In this work, we propose studying Deep Learning techniques for this purpose. Samples of Z+jet events were simulated in vacuum (pp collisions) and medium (PbPb collisions) and used to train Deep Neural Networks with the objective of discriminating between medium- and vacuum-like jets within the medium (PbPb) sample. Dedicated Convolutional Neural Networks, Dense Neural Networks and Recurrent Neural Networks were developed and trained, and their performance was studied. Our results show the potential of these techniques for the identification of jet quenching effects induced by the presence of the QGP.

Monte Carlo modeling of jets

“…To address the challenge of pinpointing jet quenching effects, various strategies, including the application of machine learning techniques, have been suggested [20][21][22] to select a jet population where quenching effects are enhanced. While these techniques proved useful, they tend to bias the jet sample towards lower where the effects of energy loss become more visible.…”

Overview: Jet quenching with machine learning

“…In relativistic heavy ion collisions, machine learning techniques are also applied in the study of jet quenching with the above motivations, including reconstructing the jet momentum [14][15][16][17][18][19], distinguishing between quenched and unquenched jets [20][21][22][23], identifying the jet energy loss [24][25][26], locating the jet creation points [27] and classifying quark and gluon jets in the heavy-ion collisions [28,29]. In the following, we will review the applications of machine learning on these topics and give an outlook for the future studies.…”

Deep learning assisted jet tomography for the study of Mach cones in QGP

“…To answer the first question, we employ a point-cloud deep neural network (DNN) with multiple hidden layers, which is powerful in pattern recognition and has been widely used in high energy nuclear physics [52][53][54][55] and jet medium interaction studies [50,51,[56][57][58][59][60][61][62][63]. We will train and validate the DNN using data from the coupled Linear Boltzmann Transport (CoLBT) and hydro model simulations [29] which combines the Linear Boltzmann Transport (LBT) model for jet parton propagation in medium with the concurrent QGP evolution given by the 3+1D CCNU-LBNL viscous (CLVisc) hydrodynamics [64][65][66].…”