Fluid-Gravity Correspondence

The fluid–gravity correspondence constructs long-wavelength, nonlinear solutions of Einstein’s equations by promoting a black brane’s temperature and velocity to slowly varying boundary fields. Radial constraint equations become hydrodynamic conservation laws; horizon regularity and boundary conditions determine constitutive coefficients. This is a derivative expansion for a particular holographic state, not an exact description at arbitrary gradients.

Required background. Ideal Relativistic Hydrodynamics supplies the ideal variables and conservation equations. Bulk Quasinormal Modes and Boundary Hydrodynamic Poles supplies the linear long-wavelength limit.

Helpful background. Hydrodynamic Frames and Constitutive Data supplies frame transformations. Hydrodynamic Attractors and Asymptotic Gradient Expansions explains why the series need not converge. Schwinger–Keldysh Effective Actions for Fluids supplies fluctuations and noise beyond the classical construction.

Boosted black branes as local equilibrium

In units with the AdS radius $L=1$, ingoing Eddington–Finkelstein coordinates put a uniform asymptotically $\mathrm{AdS}_{d+1}$ brane in the form

$$ds^2=-2u_\mu dx^\mu dr +r^2 f(br)u_\mu u_\nu dx^\mu dx^\nu +r^2P_{\mu\nu}dx^\mu dx^\nu,$$

where

$$f(br)=1-\frac{1}{(br)^d}, \qquad u_\mu u^\mu=1, \qquad P_{\mu\nu}=\eta_{\mu\nu}-u_\mu u_\nu, \qquad T=\frac{d}{4\pi b}.$$

Constant $b$ and $u^\mu$ give an exact solution. Promote them to $b(x)$ and $u^\mu(x)$ varying on a length $L_{\mathrm{grad}}\gg1/T$. The promoted zeroth-order metric is no longer exact, so expand

$$g_{ab}=g_{ab}^{(0)}[b(x),u(x)] +\epsilon g_{ab}^{(1)}+\epsilon^2g_{ab}^{(2)}+\cdots, \qquad \epsilon\sim\frac{1}{TL_{\mathrm{grad}}}.$$

At each order, the radial equations are ordinary differential equations whose sources are boundary derivatives of lower-order data.

Constraints become conservation laws

Decompose Einstein’s equations with respect to radial slices. The $r\mu$ components contain no new radial dynamical data; after lower orders are solved, they impose

$$\nabla_\mu T^{\mu\nu}=0 .$$

The remaining components determine $g_{ab}^{(n)}$. Boundary normalizability removes changes to the prescribed boundary metric, future-horizon regularity selects the causal branch, and a hydrodynamic frame fixes homogeneous zero modes that would otherwise redefine $T(x)$ and $u^\mu(x)$.

In Landau frame, $u_\mu T^{\mu\nu}=\epsilon u^\nu$, the first-order stress tensor is

$$T^{\mu\nu}=\epsilon u^\mu u^\nu-pP^{\mu\nu}+2\eta\sigma^{\mu\nu}+O(\partial^2),$$

with

$$\sigma^{\mu\nu}=P^{\mu\alpha}P^{\nu\beta} \left(\nabla_{(\alpha}u_{\beta)}-\frac{1}{d-1}P_{\alpha\beta}\nabla\!\cdot u\right).$$

For two-derivative Einstein gravity, regularity gives $\eta/s=1/(4\pi)$. Bhattacharyya et al. carried this construction through nonlinear second order and related the event-horizon area form to a boundary entropy current Bhattacharyya et al. 2008.

First-order construction as the application

Take a locally linear shear flow $u^y=\kappa x$ with constant temperature and $\lvert\kappa\rvert/T\ll1$. The only first-order tensor is $\sigma^{xy}=-\kappa/2$ in the $(+,-,\ldots,-)$ convention. Solving the tensor-channel radial equation with no boundary-metric source and a regular future horizon fixes $g_{xy}^{(1)}(r)$. Its normalizable coefficient gives

$$T^{xy}=-\eta\kappa .$$

The $ry$ constraint gives $\partial_xT^{xy}=0$ at this order. Thus the same bulk solution produces both the constitutive coefficient and its conservation equation. An outgoing or singular horizon solution would instead give the wrong dissipative sign.

Frame and gradient adversaries

Perform a first-order field redefinition

$$u^\mu\to u^\mu+\delta u^\mu, \qquad T\to T+\delta T .$$

Individual terms in $T^{\mu\nu}$ move between “ideal” and “dissipative” pieces, but the full renormalized stress tensor and shear-channel pole do not. If a reported transport coefficient changes under this transformation, the calculation has compared frame-dependent parameters rather than an invariant observable.

Next increase gradients until $\epsilon\sim1$. A smooth horizon at first order does not show that the truncated metric remains accurate: second-order terms, nonhydrodynamic quasinormal modes, and eventually caustics can compete. Compare successive orders or a full numerical solution. Regularity fixes coefficients inside the expansion; it does not prove convergence or exact late-time behavior at finite $N$.

Exercises

Why is the promoted zeroth-order metric not a solution for arbitrary $b(x)$ and $u(x)$?

Solution

Derivatives of $b$ and $u$ enter the Christoffel symbols and curvature. They create terms of order $\epsilon$ in Einstein’s equations that are absent when the parameters are constant. The correction $g^{(1)}$ cancels the dynamical terms, while the radial constraints require the boundary fields to obey hydrodynamic conservation.

The chapter overview contains the structure diagram and validity and failure diagram. They are embedded there once so that their shared chapter-level context is not repeated on every article.

For the chapter-wide comparison of assumptions, counterevidence, falsifiers, and claim ceilings, see the claim-domain table.

References

• Bhattacharyya, Sayantani, Veronika E. Hubeny, Shiraz Minwalla, and Mukund Rangamani. “Nonlinear Fluid Dynamics from Gravity.” Journal of High Energy Physics 2008, 045 (2008). DOI.