Joint Bayesian analysis of soft and high-$p_\perp$ probes yields tighter constraints on QGP properties

Abstract

To extract bulk QGP properties, we perform a joint Bayesian calibration of bulk-medium parameters using low-$\pt$ bulk and high-$\pt$ tomography within a common medium evolution. Low-$\pt$ observables are computed with \textsc{TRENTo}+\textsc{VISHNU}; temperature profiles are passed to \textsc{DREENA-A} to predict light/heavy $R_{\mathrm{AA}}(\pt)$ and $v_2(\pt)$. Gaussian-process emulation enables Hamiltonian Monte Carlo sampling of the low-$\pt$-only and joint posteriors. The low-$\pt$-only case underpredicts high-$\pt$ anisotropy; the joint calibration matches both sectors and markedly tightens bulk-parameter constraints, demonstrating the added power of high-$\pt$ data.

Figures (5)

• Figure 1: Posterior distributions of the calibration parameters ($\tau_0$, $\eta/s$, and norm) from the low-$p_{\perp}$-only Bayesian inference. Diagonal panels show marginalized 1D posteriors; off-diagonal panels show 2D joint contours.
• Figure 2: Prior (gray) and posterior predictions for the $p_{\mathrm{T}}$-integrated $h^\pm$ elliptic flow $\langle v_2\rangle$ versus centrality: low-$p_{\perp}$ only (left, blue) and joint (low-$p_{\perp}$+high-$p_{\perp}$, right, green). Red points denote the ALICE measurements ALICEv2 used in the calibration.
• Figure 3: High-$p_{\perp}$ out-of-sample test of the low-$p_{\perp}$ only posterior propagated through DREENA-A. Shown are $R_{\mathrm{AA}}$ and $v_{2}$ vs. $p_{\perp}$ for $h^\pm$ (left, 30--40%) and $D^0$ (right, 30--50%). Prior (gray) and posterior (blue) bands are compared to ATLAS ($h^\pm$) ATLASRAAATLASv2, ALICE ($D^0$$R_{\mathrm{AA}}$) ALICEDRAA, and CMS ($D^0$$v_{2}$) CMSDv2 data (red). $h^\pm$ results were computed for all four centralities; only 30--40% is shown for brevity.
• Figure 4: Same as Fig. \ref{['fig:corner_lowpt']}, but comparing overlaid posterior distributions from the low-$p_\perp$-only (blue) and joint low-$p_\perp$+high-$p_\perp$ (green) calibrations.
• Figure 5: Same as Fig. \ref{['fig:highpt_chD_lowptpost']}, but using posterior samples from the joint calibration (green).