Machine learning for high energy heavy ion collisions

Abstract

The high energy heavy ion collision is a multi-stage process that is described by complex hybrid models. The initial state fluctuations in event-by-event simulations of heavy ion collisions convert to final state correlations by collective flow and hadronic cascade. It is not easy to design final state correlations (observables) from particles in momentum space, that can help to extract useful information, such as the initial state nuclear structure, the properties of quark gluon plasma and the nuclear equation of state. Machine learning is helpful in automatic feature extraction in heavy ion collisions. This article reviews the applications, challenges and possible future developments of machine learning in heavy ion physics.