Bayesian constraints on the transport coefficients $\eta/s$ and $\zeta/s$ from spin polarization in relativisitic heavy-ion collisions

This study employs Bayesian inference to incorporate longitudinal spin polarization of $\Lambda$ hyperons alongside conventional bulk observables in Pb+Pb collisions at 5.02 TeV, demonstrating that while current uncertainties prevent a statistically significant shift in the extracted bulk viscosity, spin polarization serves as a valuable complementary probe for constraining quark-gluon plasma transport properties.

The Gist

Imagine the universe, just a fraction of a second after the Big Bang, was filled with a super-hot, super-dense soup of particles called the Quark-Gluon Plasma (QGP). Scientists smash heavy atoms (like lead) together at nearly the speed of light to recreate this soup in a lab. The big question is: How "thick" or "sticky" is this soup?

In physics, this "stickiness" is measured by something called viscosity.

• Shear viscosity ($\eta$): Think of this like honey. If you stir honey, it resists the spoon. In the QGP, this measures how much the fluid resists sliding layers past each other.
• Bulk viscosity ($\zeta$): Think of this like a sponge. If you squeeze a sponge, it resists changing its volume. In the QGP, this measures how the fluid resists expanding or compressing.

The Problem: Guessing the Recipe

For years, scientists have tried to figure out exactly how much "honey" (shear) and "sponge" (bulk) is in this cosmic soup. They use a method called Bayesian inference, which is basically a super-smart way of guessing. You start with a range of possible recipes, run a computer simulation, see how well it matches the data, and then tweak the recipe until it fits perfectly.

Until now, scientists only looked at one type of clue: how the particles fly out of the collision (their momentum). It's like trying to guess the recipe of a cake just by looking at how the crumbs scatter when you drop it. It works okay, but you might miss something important about the texture.

The New Clue: The "Spin" of the Particles

This paper introduces a new, very specific clue: Spin Polarization.

Imagine the particles in the soup (specifically, a type called $\Lambda$ hyperons) are like tiny tops. Because the collision creates a massive whirlpool (vorticity), these tops don't just spin randomly; they all try to line up in the same direction, like a school of fish turning together.

The authors realized that the way these "tops" line up (their longitudinal spin polarization) is extremely sensitive to the "sponge-like" resistance (bulk viscosity) of the soup. It's a different kind of clue than the flying crumbs.

What They Did

The team built a massive computer model of a lead-lead collision.

1. The Simulator: They created a "virtual lab" where they could change the viscosity settings (the recipe) and run the collision millions of times.
2. The Emulator: Since running the full physics simulation takes forever, they built a "smart shortcut" (a Gaussian Process emulator) that could predict the results instantly.
3. The Test: They ran their Bayesian analysis twice:
   • Test A: Using only the old clues (flying particles).
   • Test B: Using the old clues PLUS the new spin clues (how the tops lined up).

The Results: A Surprising Shift

Here is what they found, explained simply:

• The "Honey" (Shear Viscosity) didn't change much.
  The old clues were already very good at telling them how "honey-like" the soup was. Adding the spin clue didn't change their guess. The soup is still very runny, almost like a perfect fluid.

• The "Sponge" (Bulk Viscosity) changed a lot.
  When they added the spin clue, their guess for the "sponge-ness" of the soup doubled.

  • Without the spin clue: They thought the soup was fairly easy to compress.
  • With the spin clue: They realized the soup is actually much harder to compress (more "sponge-like").

Why This Matters

The paper concludes that the "spin" of the particles is a secret decoder ring for the bulk viscosity. If you only look at how particles fly, you might think the soup is less "spongy" than it really is.

The authors argue that to get the true recipe of the Quark-Gluon Plasma, scientists must stop ignoring the spin. It provides a unique, complementary view that helps pin down the "sponge" properties of the universe's most perfect fluid.

In short: They used a new type of evidence (spinning tops) to fix a blind spot in their understanding. The soup is still a perfect fluid, but it turns out to be much more "spongy" than they previously thought.

Technical Summary

Technical Summary: Bayesian Constraints on Transport Coefficients $\eta/s$ and $\zeta/s$ from Spin Polarization

Problem Statement
Relativistic hydrodynamics has successfully described the quark-gluon plasma (QGP) as a near-perfect fluid, yet the transport coefficients governing its evolution—specifically the shear viscosity to entropy density ratio ($\eta/s$) and the bulk viscosity to entropy density ratio ($\zeta/s$)—remain difficult to compute from first principles due to the non-perturbative nature of Quantum Chromodynamics (QCD). While Bayesian inference frameworks have been widely adopted to constrain these coefficients using experimental data, previous analyses have relied almost exclusively on bulk hadronic observables constructed from momentum degrees of freedom (e.g., multiplicities, mean transverse momenta, and anisotropic flow). These studies have not incorporated spin-related measurements, despite evidence that the QGP is the most vortical fluid observed and that $\Lambda$ hyperon spin polarization is sensitive to the medium's space-time structure and vorticity. Recent theoretical developments suggest that longitudinal spin polarization ($P_z$) is particularly sensitive to bulk viscosity and initial flow structures, motivating the need to include these observables in a unified Bayesian analysis to provide independent constraints on QGP transport properties.

Methodology
The authors perform a systematic Bayesian analysis of Pb+Pb collisions at $\sqrt{s_{NN}} = 5.02$ TeV to constrain $\eta/s$ and $\zeta/s$. The study employs a (3+1)-dimensional viscous hydrodynamic framework coupled with a hadronic afterburner (SMASH) and a spin-hydrodynamic formalism to calculate $\Lambda$ hyperon polarization.

• Model Setup: The hydrodynamic evolution is governed by energy-momentum and net baryon number conservation in the Landau frame. The shear and bulk viscous pressures are treated using second-order Israel-Stewart-like equations. The transport coefficients are parameterized as temperature-dependent functions with three free parameters each for $\eta/s(T)$ and $\zeta/s(T)$. The equation of state is based on lattice-QCD results (NEOS-BQS).
• Initial Conditions: To render the Bayesian analysis computationally feasible, the authors utilize event-averaged initial profiles rather than full event-by-event simulations. Initial conditions are generated using the TRENTo model, smoothed by a nucleon width parameter, and averaged over 5,000 events per centrality class. Pre-equilibrium dynamics are neglected, with hydrodynamics initialized at a proper time $\tau_0$.
• Spin Polarization Calculation: The spin polarization vector is computed on the particlization hypersurface, incorporating contributions from thermal vorticity and thermal shear, following the prescription of Becattini et al. [41, 42]. The analysis focuses on the longitudinal component ($P_z$) and adopts an isothermal approximation for the switching hypersurface.
• Bayesian Inference Framework:
  • Emulator: Due to the computational cost of full hydrodynamic simulations, a Gaussian Process Regressor (GPR) emulator is trained on a design set of 793 model evaluations generated via Latin Hypercube Sampling (LHS) across a 13-dimensional parameter space.
  • Dimensionality Reduction: Principal Component Analysis (PCA) is applied to the model outputs to reduce the observable space to three principal components, capturing over 97% of the variance.
  • Likelihood and Sampling: A multivariate Gaussian likelihood is constructed in the principal component space, incorporating emulator, experimental, and truncation uncertainties. Posterior distributions are sampled using a parallel-tempered Markov Chain Monte Carlo (MCMC) algorithm.
  • Data: The analysis includes nine sets of experimental data across seven centrality classes: charged particle pseudorapidity density, identified particle rapidity densities, mean transverse momenta, elliptic flow ($v_2$), and the longitudinal spin polarization ($P_z$) of $\Lambda$ hyperons.

Key Contributions
This work presents the first Bayesian constraints on QGP transport coefficients that explicitly incorporate longitudinal spin polarization observables alongside conventional bulk measurements. The study demonstrates a rigorous methodology for integrating spin degrees of freedom into global hydrodynamic parameter extractions, utilizing a Gaussian process emulator to handle the computational demands of (3+1)D spin-hydrodynamics.

Results

• Shear Viscosity ($\eta/s$): The inclusion of spin polarization data does not significantly alter the extracted shear viscosity. The posterior distributions for $\eta/s(T)$ remain largely unchanged compared to analyses using only bulk observables, indicating that conventional hadronic data already provide strong constraints on shear transport.
• Bulk Viscosity ($\zeta/s$): The inclusion of $P_z$ data shifts the posterior distribution of the bulk viscosity toward larger values. Specifically, the median value of $\zeta/s(T)$ increases by approximately a factor of two when spin polarization is included, although the Maximum A Posteriori (MAP) estimates show only a modest upward shift.
• Statistical Significance: While the shift in the bulk viscosity median is notable, the 68% credible intervals of the analyses with and without spin polarization retain significant overlap (particularly at temperatures $T \gtrsim 300$ MeV). Consequently, the authors state that the current uncertainties do not allow for a statistically significant separation at the 68% credibility level.
• Model Consistency: The parameter set extracted from the analysis including spin polarization successfully reproduces the experimentally observed sign of the longitudinal polarization $P_z$, whereas the parameter set derived without spin data fails to do so. Both parameter sets, however, provide a satisfactory description of conventional bulk observables (multiplicities, $p_T$ spectra, and $v_2$).
• Initial Conditions: Parameters related to initial conditions and hydrodynamic initialization (e.g., $\tau_0$, nucleon width smoothing) exhibit broad, multi-modal posterior distributions, indicating they are weakly constrained by the current set of observables.

Significance and Claims
The paper claims that spin polarization measurements provide complementary and nontrivial information for the quantitative extraction of QGP transport properties, particularly for bulk viscosity. The results suggest that longitudinal spin polarization is sensitive to the space-time structure of the expansion dynamics, which is directly influenced by bulk viscosity. While the current data uncertainties prevent a definitive statistical separation of the bulk viscosity values, the shift in the posterior median and the successful reproduction of the $P_z$ sign demonstrate that spin observables should be incorporated into comprehensive Bayesian extractions. The authors conclude that future systematic analyses including polarization observables are necessary to further refine constraints on the dissipative properties of the quark-gluon plasma.