Bayesian Model Selection in (3+1)D Bayesian Analysis of RHIC BES data and Predictions for Upcoming RHIC BES measurements

We present our work on Bayesian model selection to introduce collision energy dependence in additional model parameters in our (3+1)D framework 3DGlauber + Hydrodynamics + UrQMD to analyze RHIC BES data. Using the Bayes factor, we identify that only the model setup with collision energy dependence on initial transverse and longitudinal hotspot width is moderately favored. We also explore how the addition of observables such as normalized charged hadron p_T-fluctuations, p_T-differential elliptic flow, additional particle yields and mean p_T progressively influences model predictions and the posterior probability distributions to explore how different experimental observables constrain different model parameters. We sampled a few parameter sets from the posterior distribution to make full model predictions. We made model predictions for rapidity-dependent elliptic flow, longitudinal flow decoorelation, identified particle v₀(p_T) for Au+Au systems at different RHIC BES energies alongside predictions for elliptic flow for small systems. The prediction uncertainty from our analysis is quantified as twice the standard deviation of the observable across the parameter sets. Our results are important for making model to data comparisons for the upcoming RHIC BES measurements. Our work also presents an important contribution towards exploring (3+1)D phenomenology for different collision systems.