Quantifying the Quark Gluon Plasma

Abstract

The study of heavy-ion collisions presents a challenge to both theoretical and experimental nuclear physics. Due to the extremely short (10 −23 s) lifetime and small size (10 −14 m) of the collision system, disentangling information provided by experimental observables and progress in physical insight requires the careful application of plausible reasoning.I apply a program of statistical methodologies, primarily Bayesian, to quantify properties of the medium in specific models, as well as compare and criticize differing models of the system. Of particular interest are estimations of the specific shear and bulk viscosities, where we find that information carried by the experimental data is still limited. In particular we find a large sensitivity to prior assumptions at high temperatures. Moreover, sensitivities to model assumptions are present at low temperatures, and this source of model uncertainty is propagated with model averaging and model mixing.

Bayesian quantification of strongly-interacting matter with color glass condensate initial conditions

“…The tension involving dE T /dη is not new to this work and was also seen in [43]. This suggests that it is a feature independent of the pre-equilibrium stage and potentially a feature of either the specifics of the hadronic chemistry or -less likely -a feature of 2+1D hydrodynamics.…”

“…As is often the case in fields with a plenitude of data, Bayesian model averaging remains the current state-ofthe-art in heavy-ion collisions for leveraging the information in multiple models to best constrain the physical understanding of strongly-interacting matter without over-fitting. This is only the second study in this field, following [29] and elaborated in [43], to utilize BMA for improving uncertainty quantification and has further demonstrated its importance. Further sources of unquantified uncertainty still exist in heavy ion collisions, usually at the interface between models at each stage in the evolution of the fireball, but how to incorporate such interface effects in BMA is not yet clear.…”

“…It does this by mapping inputs to outputs and learning the functional relationship between them rather than attempting to produce a coarse version of the intermediary physics. These methods have had success in heavy ion physics [25,[28][29][30][31][39][40][41][42][43][44].…”

“…It has been proposed to avoid using a parametrization, which correlates values of the viscosity at different temperatures [43], an idea that has recently been explored in other context in heavy ion collisions [86]. This is beyond the scope of this work.…”

“…BMA was first used in heavy ion collisions to perform model averaging of the transport coefficients and later for model averaging of non-viscous parameters [29,43]. The BMA viscous posteriors are shown in Fig.…”