Accessing the Equation of State from Heavy-Ion Data: A Deep-learning Approach*

One of the main goals of the heavy-ion collision research programme is a deeper understanding of the quark gluon plasma (QGP) – the state of matter consisting of a free collection of quarks and gluons. We do not access it directly, however, but rather probe its properties via the kinematic data of particles detected post-hadronization. A major difficulty, then, is extracting the thermodynamic properties of the QGP based on kinematic data. Recently Pang, et al. were able to use a convolutional neural network (CNN)  to determine said thermodynamic properties from hydrodynamic simulations of final-state particle distributions in transverse momentum (pT) and azimuthal angle (ɸ). In determining from kinematic data whether the equation of state is first-order or a crossover, one can probe the phase diagram of the QGP in search of the QCD critical point that separates the first-order and crossover regimes. We attempt to extend this work, considering novel architectures, sources of data, and methods of network interpretability.