Abstract
We explain why the maximal positive Lyapunov exponent of classical $SU(N)$ gauge theory coincides with (twice) the damping rate of a plasmon at rest in the leading order of thermal gauge theory. [This is a substantially revised and expanded version of the manuscript.]
Introduction
In the study of high-temperature non-Abelian gauge theories, such as those describing the quark-gluon plasma, two distinct fields of physics often intersect: nonlinear dynamics and thermal field theory. One of the most intriguing observations from numerical lattice simulations in the early 1990s was the discovery that classical Yang-Mills fields are inherently chaotic. This chaos is characterized by the maximal Lyapunov exponent ($\lambda_0$), which measures the exponential rate at which two slightly different field configurations diverge over time.
Simultaneously, thermal field theory provides a framework to calculate the damping rate ($\gamma_0$) of collective excitations, known as plasmons. Curiously, numerical results for $SU(2)$ and $SU(3)$ gauge theories consistently showed that $\lambda_0 \approx 2\gamma_0$. This paper provides the theoretical derivation to explain this relationship, demonstrating that the "chaoticity" of the classical field effectively serves as a thermal bath for its own excitations.
Methodology
To reconcile deterministic chaos with thermal dissipation, the authors employ a multi-step analytical approach:
- Linearized Perturbation Analysis: Starting with the classical Yang-Mills equations, the authors introduce a small perturbation $a_\mu$ and track its evolution. This allows them to determine whether the perturbation grows (indicating chaos) or oscillates.
- The Ergodic Limit: Because the system is highly chaotic, it is assumed to be ergodic. This allows the researchers to replace long-time averages of the field's evolution with canonical (thermal) ensemble averages. In this "self-thermalizing" environment, the chaotic dynamics provide the same scattering mechanism for collective modes as thermal fluctuations do in a quantum plasma.
- Stochastic Frequency Modeling: A key innovation is modeling the squared frequency of field fluctuations ($\Omega^2$) as a Gaussian-distributed stochastic variable. In a chaotic system, the "restoring force" can occasionally become negative, leading to exponential growth rather than stable oscillation.
- Spectral Function Analysis: By using Laplace transforms—which are better suited for initial value problems in forward time than standard Fourier transforms—the authors derive a four-pole spectral function. The poles in the complex frequency plane represent both the growth (Lyapunov) and decay (damping) rates.
Explaining the Factor of Two
The most direct result of the paper is the formal proof of the identity: $$\lambda_0 = 2\gamma_0$$
The origin of the factor of two lies in the definition of the distance measure used in numerical simulations. Lattice studies typically measure the divergence of trajectories using the magnetic energy density, which is quadratic in the field variables ($D \propto a^2$). If the field amplitude $a$ grows at a rate $e^{\gamma t}$, the energy density—and thus the distance measure—grows at $e^{2\gamma t}$.
This relationship holds specifically for a plasmon at rest ($k=0$). The authors show that the leading-order damping rate $\gamma_0$ is actually a purely classical quantity. While often derived via quantum field theory (using Hard Thermal Loop resummation), the damping rate survives the $\hbar \to 0$ limit, confirming its classical nature.
Key Results
The paper confirms the following numerical relationships, aligning theoretical predictions with previous lattice simulation data:
- For $SU(2)$: $\lambda_0 \approx 0.34 g^2 T$ and $\gamma_0 \approx 0.176 g^2 T$.
- For $SU(3)$: $\lambda_0 \approx 0.53 g^2 T$ and $\gamma_0 \approx 0.264 g^2 T$.
These findings suggest that the maximal Lyapunov exponent is not merely a mathematical curiosity of lattice simulations but a fundamental physical property. It reflects the rate at which the system loses memory of its initial state, which is intrinsically linked to the rate at which collective excitations dissipate their energy into the thermal background.
Limitations and Future Outlook
While the identity $\lambda_0 = 2\gamma_0$ is robust for $k=0$, several questions remain. The authors acknowledge that the ergodic assumption, while supported by data, remains a heuristic step. Furthermore, the relationship for non-zero momentum ($k \neq 0$) remains a conjecture. In classical theories, the plasmon mass diverges in the continuum limit ($a \to 0$), yet the authors argue that the damping rate (and the Lyapunov exponent) remains finite and independent of this divergence. Future work is needed to determine if the entire spectrum of Lyapunov exponents corresponds to the damping rates of excitations across different momenta.
Summary updated