Neural network bridges lattice QCD and heavy-ion collision data

Heavy-ion collisions probe matter at extreme densities, but the equation of state—how pressure, energy, and density relate—remains poorly constrained in the high-baryon regime. This work uses a neural network trained on lattice QCD data and hadron thermodynamics to extrapolate into the unmeasured region targeted by RHIC, FAIR, and CBM experiments. The network enforces thermodynamic consistency throughout phase space, enabling realistic simulations of collisions from low to high density.