Bayesian inference and jet quenching (2507.22288v1)

Abstract: These proceedings review the application of Bayesian inference to high momentum transfer probes of the quark--gluon plasma (QGP). Bayesian inference techniques are introduced, highlighting critical components to consider when comparing analyses. Recent calibrations using hadron observables are described, illustrating the importance of the choice of parametrization. Additional recent analyses that characterize the impact of the inclusion of jet observables, as well as soft-hard correlations, are reviewed. Finally, lessons learned from these analyses and important questions for the future are highlighted.

• The paper demonstrates that Bayesian inference systematically calibrates QGP models by integrating multiple observables from jet quenching data.
• The methodology employs joint calibration of soft and hard sector parameters, enhancing constraints on the jet transport coefficient.
• Results from RHIC and LHC data validate model consistency and set a foundation for future refinements using advanced computational methods.

Bayesian Inference and Jet Quenching

Introduction to Bayesian Inference in Jet Quenching

The application of Bayesian inference to jet quenching in high-energy physics provides significant insights into the complex dynamics of the quark-gluon plasma (QGP). Jet quenching measurements from RHIC and LHC serve as probes for the QGP, but traditional single-observable analyses are limited in discriminating between various theoretical models. Bayesian inference offers a robust framework for comparing model predictions across multiple observables, thus providing a systematic approach to parameter estimation within QGP models.

Bayesian inference operates via Bayes' theorem, which relates model parameters ($\vec{\theta}$) to experimental data ($\vec{x}$). The posterior distribution obtained, $P(\vec{\theta} | \vec{x})$, encodes the probability of parameter values given the observed data, allowing researchers to assess model consistency and identify areas needing improvement.

Figure 1: $extracted for a physics-inspired (dotted fill)~\cite{JETSCAPE:2021ehl} and an information field~\cite{Xie:2022ght,Xie:2022fak} (red)$ parameterization calibrated on hadron yield modification data.

Calibrations with Hadron Measurements

Recent advancements in Bayesian analyses use hadron yield modification measurements to calibrate QGP models. The JETSCAPE Collaboration and others have employed Bayesian frameworks to integrate inclusive hadron, dihadron, and $\gamma$-hadron data from both RHIC and LHC, utilizing different approaches to parametrizing $\hat{q}$, the jet transport coefficient. These analyses demonstrate consistency in posterior distributions across various parametric choices.

Figure 2: $$extracted from differential selections of RHIC data measured at$$ = 200 GeV (left) and LHC data measured at $$= 5.02 TeV (right) as a function of centrality selection. All calibrations are consistent at fixed$$, with improved constraints for more data.

Soft-Hard Calibrations

Inclusivity in calibrations, considering both soft and hard sector parameters, is key for understanding QGP dynamics comprehensively. A calibration utilizing both low-$p_$ and high-$p_$ observables revealed shifts in the posterior distribution's most probable values, indicating correlations between soft and hard sector dynamics. This highlights the need for joint calibrations in future analyses.

Figure 3: Parameter posterior distributions from the calibration of a subset of the soft-sector parameters in the DREENA-A model for low-$p_$ (blue) and $p_$-inclusive observables (green).

Calibrations with Inclusive Hadron and Jet Measurements

The integration of jet measurements with hadron data in Bayesian inference has refined the understanding of QGP parameters. The calibration of models such as LIDO and JETSCAPE demonstrates consistent $\hat{q}$ values, though some discrepancies arise due to data selection and model parametrization choices.

Figure 4: $$posterior distributions as a function of T extracted by the calibration of the LIDO model to selected hadron and jet measurements.* (Figure 5) *Figure 5: (left)$$ posterior distribution extracted for hadron-only and jet-only observables as a function of T. (Right) $$ posterior distribution extracted for jet-only observables (blue), hadron-only observables (orange), and selections of hadron data for p_ > 10 \GeVc{}.

Outlook

Bayesian inference is pivotal for advancing QGP research, facilitating model comparisons and driving improvements in theoretical formulations. Future directions include high-precision hadron and jet substructure measurements, leveraging machine learning for computational efficiency, and systematic inclusion of prior information in Bayesian analyses. The goal is to enable comprehensive model discrimination using RHIC and LHC data, providing deeper insights into the microscopic properties of QGP.

In the coming years, the field will continue to explore computational methods that allow optimal calibration between soft and hard processes, enhancing predictions across observable spaces. By addressing theoretical uncertainties and advancing experimental methodologies, Bayesian analyses will remain integral in exploring the multifaceted interactions within QGP.