22–27 Mar 2026
US/Pacific timezone

Extract the QCD speed of sound in the presence of quantum fluctuations

24 Mar 2026, 15:15
20m
PAB 1434 (Physics and Astronomy Building)

PAB 1434

Physics and Astronomy Building

Oral Presentation Parallel V: Phase Structure

Speakers

Li Yan (Fudan University) Yushan Mu (Fudan University)

Description

It has recently been realized that in the ultra-central heavy-ion collisions, mean transverse momentum of hadrons contains the information of the fundamental thermodynamic properties of quark-gluon plasma (QGP). In particular, in nucleus-nucleus collisions, the linear correlation between the mean transverse momentum and the charged multiplicity is attributed to the QCD speed of sound, which promotes both theoretical and experimental investigations. However, in realistic collisions, these studies suffer from the contamination of fluctuations, especially the quantum fluctuations from the initial state, which bias the extracted value. Traditional analyses struggle to separate this fluctuating background from the genuine thermodynamic signal.

In this talk, we present a systematic subtraction scheme to resolve this issue. In a thermalized QGP, the quantum fluctuations $\delta$ are independent from the thermodynamic responce and vary randomly across events. According to the Central Limit Theorem, the distribution of $\delta$ should approach Gaussianity allowing us to extract the physical speed of sound statistically even in the presence of these fluctuations. Crucially, this approach can also serve as a direct probe of QGP thermalization. In a non-thermalized system, the distribution of $\delta$ deviates from Gaussianity and the extracted value of speed of sound is non-physical. The deviations from thermalization can be quantified by the standardized kurtosis $\kappa_4$ of $\delta$.

Validated by the event-by-event hydrodynamic simulations, the extracted speed of sound in our framework is successfully consistant with the predictions from lattice QCD, from large to small collision systems. To enhance statistical robustness, we also employ AI-powered diffusion model for data augmentation, which further strengthen the reliability of our results.

Authors

Li Yan (Fudan University) Prof. Xu-Guang Huang Yushan Mu (Fudan University)

Presentation materials