Hydrodynamic Poles and Transport

Hydrodynamic poles are the quasinormal modes forced to approach zero frequency by conserved densities. Their small-momentum coefficients encode diffusion, viscosity, sound speed, and attenuation. In many holographic plasmas the dispersion series has a finite analytic domain, often bounded by a collision between the hydrodynamic pole and a gapped quasinormal mode at complex momentum. The nearest actual singularity, not an arbitrary real-momentum accuracy threshold, determines the series radius.

Required background. Gauge Theory and String-Theory Preparation provides stress-tensor and linear-response conventions; Quantum Information provides retarded-response language.

Helpful background. Quantum Gravity supplies the classical bulk regime, while Black-Hole Perturbation Theory provides gauge-invariant channels and quasinormal boundary conditions.

Conservation fixes the leading poles

Choose the Landau frame and perturb a neutral relativistic fluid around equilibrium. In a transverse shear channel, momentum conservation gives

$$\omega_{\rm shear}(k)=-iD_\eta k^2+O(k^4), \qquad D_\eta=\frac{\eta}{\epsilon+p}.$$

In the longitudinal channel,

$$\omega_{\rm sound}^{\pm}(k)= \pm c_s k-\frac{i}{2}\Gamma_s k^2+O(k^3),$$

where $c_s^2=(\partial p/\partial\epsilon)$ in the stated thermodynamic ensemble and, in three spatial dimensions, $\Gamma_s=(4\eta/3+\zeta)/(\epsilon+p)$. These statements follow from constitutive relations plus Ward identities; gravity computes the transport coefficients by solving the ingoing bulk perturbation problem. Kovtun and Starinets 2005, §§3.2–4.2, pp. 10–23 construct the gauge-invariant shear and sound variables and obtain the corresponding holographic poles.

The order of limits matters. A diffusion constant is extracted with $\omega$ scaling as $k^2$; a static susceptibility takes $\omega\to0$ before $k\to0$. Interchanging them can select a different response. The boundary conditions are ingoing at the future horizon and source-normalized at the AdS boundary, with Fourier convention $e^{-i\omega t}$.

The independent Kubo and Ward checks

The same shear viscosity follows from the zero-momentum Kubo formula

$$\eta=-\lim_{\omega\to0}\frac{1}{\omega} \operatorname{Im}G_R^{T^{xy}T^{xy}}(\omega,\mathbf{0}).$$

Agreement between this result and the shear-pole coefficient is an independent check because the two calculations use different correlator components and different limits. Ward identities then relate energy and momentum correlators and fix the relative pole residues. A frequency with the correct dispersion but a residue violating the Ward identity signals a gauge, counterterm, or normalization error.

For a two-derivative conformal black brane, $\epsilon+p=sT$ and $\eta/s=1/(4\pi)$, so $D_\eta=1/(4\pi T)$. This is a benchmark in that regime, not an immutable bound: higher-derivative and finite-coupling corrections can change it.

Sound supplies a second cross-check. The equilibrium equation of state fixes $c_s$, while the imaginary part of the pole tests viscous attenuation. In a conformal plasma in three spatial dimensions, tracelessness gives $\epsilon=3p$, hence $c_s^2=1/3$ and $\zeta=0$. Recovering both numbers from the bulk sound channel checks the holographic renormalization and constraint equations simultaneously. Failure of either equality can be physical when conformal symmetry is broken, so the equation of state must be checked before using this benchmark.

Pole collision and convergence radius

Write a channel’s exact quasinormal spectrum as a spectral curve

$$P(\omega,k^2)=0.$$

The hydrodynamic dispersion is the branch satisfying $\omega(0)=0$. A candidate multiple root occurs where

$$P(\omega_c,k_c^2)=0, \qquad \partial_\omega P(\omega_c,k_c^2)=0.$$

The implicit-function theorem no longer guarantees an analytic root there, but these equations alone do not prove a branch point: two analytic branches can cross. With $q=k^2$, a generic square-root collision also has $P_{\omega\omega}\ne0$ and $P_q\ne0$, giving

$$\omega-\omega_c\simeq \pm\sqrt{-\frac{2P_q}{P_{\omega\omega}}(q-q_c)}.$$

The nearest actual singularity on the branch being followed fixes its Taylor-series radius. Grozdanov, Kovtun, Starinets, and Tadić 2019, §§2.1–2.3 and 4.6, pp. 8–15 and 31–32 formulate this spectral-curve analysis and compute holographic examples. The limiting collision need not occur at real momentum and need not involve the least damped gapped mode at $k=0$. Sound already has $\omega\sim\pm c_s\sqrt q$ at the origin; its two sound branches are expanded in $k$, not as a single Taylor series in $q$.

A useful solvable model is the relaxation equation $\tau\partial_t^2 n+\partial_t n-D\nabla^2n=0$, with $\tau,D\gt0$. Its roots are

$$\omega_\pm(k)=\frac{-i\pm i\sqrt{1-4\tau Dk^2}}{2\tau}.$$

The plus branch begins at $-iDk^2$, while the minus branch begins at $-i/\tau$. They meet at $k^2=1/(4\tau D)$, explicitly displaying a finite dispersion radius. This is a teaching model, not an exact holographic spectrum. The convergent linear dispersion series also differs from a nonlinear late-time hydrodynamic expansion, which can be asymptotic.

A practical check compares successive truncated series with the full numerical pole. Accuracy should improve inside $\lvert k\rvert\lt\lvert k_c\rvert$ and eventually deteriorate outside. Merely fitting a few real momenta cannot locate a complex branch point reliably.

Validity and physical interpretation

Hydrodynamics is controlled by small gradients compared with the microscopic relaxation scale, not by weak interactions. A strongly coupled plasma can have excellent hydrodynamics because conservation isolates slow modes. Conversely, a large wavelength does not guarantee that a low-order truncation is accurate near a complex-plane obstruction.

Pole collisions describe analytic continuation of linear response. They do not by themselves determine nonlinear hydrodynamization or guarantee that the same scale controls every observable. Broken translations, conserved charges, magnetic fields, or critical modes alter the channel content and can move or multiply the obstructions. Each calculation must state the conserved sector, susceptibility, hydrodynamic frame, and boundary quantization.

Exercises

Derive the shear diffusion pole from momentum conservation and explain how a collision condition determines the radius of its small-momentum series.

Solution

For a transverse velocity perturbation, $\delta T^{0y}=(\epsilon+p)\delta u_y$ and $\delta T^{xy}=-\eta\,\partial_x\delta u_y$. Conservation gives $-i\omega(\epsilon+p)\delta u_y+\eta k^2\delta u_y=0$, hence $\omega=-i\eta k^2/(\epsilon+p)$. A simple spectral root has a locally analytic continuation. At $P=P_\omega=0$, check the higher derivatives and the branch actually followed: the nondegenerate square-root case above gives a genuine obstruction, whereas $P=(\omega-q)(\omega+q)$ has analytic crossing branches $\omega=\pm q$. The closest genuine singularity fixes the radius; the mere failure of a theorem’s sufficient condition does not.

References

• Grozdanov, Sašo, Pavel K. Kovtun, Andrei O. Starinets, and Petar Tadić. “The Complex Life of Hydrodynamic Modes.” Journal of High Energy Physics 11 (2019) 097. arXiv:1904.12862.
• Kovtun, Pavel K., and Andrei O. Starinets. “Quasinormal Modes and Holography.” Physical Review D 72 (2005) 086009. arXiv:hep-th/0506184.