Using Machine Learning to Analyze The Effects of Quark-Gluon Plasma

In nuclear physics, a "jet" is defined as a cone of hadrons and other particles that are formed from the hadronization of a quark or gluon. Furthermore, Quark-Gluon Plasma(QGP) is a deconfined state of quarks and gluons predicted by the quantum chromodynamics (QCD) theory. QGP can be formed in heavy ion collisions at RHIC and LHC. A key signature of QGP is "jet quenching" where high energy partons showering into a spray of hadrons travel through the QGP and lose energy. Previously, this jet-by-jet quenching effect has been studied with jet substructure modifications and a machine learning approach.To build on this, in my study, I simulated pp and PbPb collisions and then compared the resulting jet substructure observables. This was done by creating a Neural Network(NN) algorithm that analyzed and classified the simulation results. My primary goal was to optimize the NN by: fine-tuning the hyper parameters, increasing the batch size of jets used during analysis, and calibrating the classification threshold. Through this approach to implement machine learning into our analysis, the following study works towards gathering a more comprehensive understanding of the performance and limitations of NNs in jet quenching identification.