Underlying Event Fluctuations and Jet Background in √sNN = 200 GeV Au+Au Collisions Using the sPHENIX Detector

At the Relativistic Heavy Ion Collider at Brookhaven National Laboratory, heavy ions are collided at ultra-relativistic speeds. The motivation for these collisions is to study an extreme form of Quantum Chromodynamical matter called Quark-Gluon Plasma (QGP). The QGP can be studied using collimated sprays of particles that are products of initial hard-scatterings between colliding partons in the moments prior to the production of QGP. These collimated sprays of particles are called jets. Comparing how jets are modified in heavy ion collision systems to that of proton-proton collisions, where QGP production is not expected, provides a useful tool to probe the strongly interacting medium and test predictions of Quantum Chromodynamics.

Measurements of jets in heavy ion collisions must contend with the high entropy and the presence of a large, fluctuating background of soft particles unrelated to the hard-scattering event. This underlying event and its fluctuations must be well understood in order to correct for it in the most challenging jet kinematic regimes. Jet momentum reconstruction for low momenta jets in heavy ion collisions at the Relativistic Heavy Ion Collider is particularly difficult due to the low initial momentum transfer of the hard-scatterings, and the large fraction of gluon-initiated jets.

In this analysis, the underlying event in Au+Au collisions at a center of mass energy of 200 GeV is measured using the sPHENIX detector. The spatial and multiplicity correlations for underlying event fluctuations are presented and the sources of underlying event fluctuations are determined and quantified. The fluctuations of the underlying event are found to be primarily driven by statistical fluctuations in particle multiplicity and non-stochastic contributions to the underlying event are found to be primarily due to initial geometric anisotropies in the colliding nuclei.

Direct comparisons between jet background subtraction methods are presented based on each jet background subtraction method's ability to suppress underlying event fluctuations. This is the first direct comparison between methods used by different experiments for jet background reconstruction. Machine learning applications to jet background subtraction are presented and their applicability to data is discussed. Interpretable machine learning techniques are applied to previously published jet background subtraction methods and a physics motivation is presented based on learned expressions. We apply a new jet background subtraction method, based on jet constituent multiplicity, to data for the first time. Our findings provide guidance for future jet measurements at sPHENIX.