Exploring the QCD phase transition in relativistic heavy-ion collisions with fluctuations of conserved charges and machine learning

点击翻译按钮获取中文摘要

Properties of strongly interacting matter at extreme densities and temperatures are important for our understanding of the fundamental constituents of matter. The equation of state at high densities and moderate temperatures is the most important property of nuclear matter that is also relevant in cosmic events such as the recently observed neutron star mergers. At zero baryon chemical potential, a smooth cross-over has been established by lattice QCD and is consistent with experimental data in ultra-relativistic heavy-ion reactions at RHIC (Relativistic Heavy Ion Collider) and LHC (Large Hadron Collider). At finite net baryon density the sign problem prevents first principle calculations and one can only resort to an experimental exploration of the phase diagram of strongly interacting matter. Fluctuations of conserved charges are promising observables for a first order phase transition or critical endpoint in the QCD phase diagram. In this proposal, we plan to carry out a close experiment-theory collaboration to identify signatures of the phase transition between the quark-gluon plasma and the hadron gas. One critical task is to scrutinize measurements of higher moments of conserved charges and quantify background effects. Within dedicated transport and hydrodynamic calculations, multi-particle efficiencies, kinematic cuts, conservation laws and baryon transport are going to be investigated in detail. Volume fluctuations affected by the centrality selection are going to be addressed as well. The baryon stopping that determines the amount of net baryon density in the system will be investigated in a hadron-string approach. The goal is to clarify ambiguities in existing measurements by extracting the amount of the fluctuations associated with the physics of the critical endpoint and provide guidance for future experiments. In addition, machine learning techniques will be applied to identify new sensitive observables. Hybrid calculations with different equations of state will be used to train a deep convolutional neural network to identify the EoS from simulated and real experimental data. For this purpose, the highly efficient GPU based algorithm CLVisc for 3-dimensional viscous hydrodynamics is going to be extended to finite net baryon densities. By combining the complementary expertise of the Chinese and German applicants, the proposed exploration will provide insights on new routes to a better understanding of the QCD phase diagram.