Near-Horizon Extreme Kerr Geometry

The near-horizon extreme Kerr geometry, usually abbreviated NHEK, is the exact vacuum spacetime obtained by zooming into the degenerate horizon of extremal Kerr. The direct scaling gives a Poincaré throat patch whose global extension is a complete spacetime with enhanced $\mathrm{SL}(2,\mathbb R)\times\mathrm U(1)$ symmetry. It preserves the horizon area and rotation but not asymptotic flatness. The geometry is not a second exterior region of Kerr; it is a distinct scaling limit.

Helpful background. Exact Solutions of Einstein’s Equations supplies the Kerr metric and Boyer–Lindquist coordinates; Causality and Global Structure supplies patch and extension language; The Einstein Equation supplies the vacuum check; Black-Hole Perturbation Theory treats fields on Kerr and NHEK; Black-Hole Geometry and Mechanics supplies extremal horizon generators and the near-horizon limit.

From extremal Kerr to the throat

Take Kerr with positive angular momentum and

$$\begin{aligned} a&=GM,\\ J&=GM^2,\\ r_+&=GM. \end{aligned}$$

In Boyer–Lindquist coordinates $(\widetilde t,\widetilde r,\theta, \widetilde\phi)$, introduce corotating scaling coordinates by

$$\begin{aligned} \widetilde r&=GM+\lambda r,\\ \widetilde t&=\frac{t}{\lambda},\\ \widetilde\phi&=\phi+\frac{t}{2GM\lambda}, \end{aligned}$$

and hold $(t,r,\theta,\phi)$ fixed as $\lambda\to0$. The shift subtracts the horizon angular velocity $\Omega_H=1/(2GM)$; without it, the angular term would not have a finite limit. Bardeen and Horowitz carry out this substitution explicitly in Bardeen and Horowitz 1999, §2, pp. 3–4, Eqs. (2.1)–(2.6).

In particular, $\widetilde\phi-\Omega_H\widetilde t=\phi$. Fixed scaled $r$ samples $\widetilde r-GM=\lambda r\to0$, while a fixed positive parent offset $\rho_*\gt0$ appears at $r=\rho_*/\lambda\to\infty$. This is the Kerr realization of the general near-horizon coordinate blow-up, not a $\lambda=0$ coordinate chart on the parent exterior.

Take $\lambda$ to be dimensionless, so both $t$ and $r$ have dimensions of length. Introduce the dimensionless throat coordinates

$$T=\frac{t}{2GM}, \qquad R=\frac{r}{GM}, \qquad \Phi=\phi.$$

Then $R\,dT=r\,dt/[2(GM)^2]$, which fixes the normalization of the azimuthal fibration rather than leaving the rescaling implicit. Define

$$\begin{aligned} L^2&=2GJ=2(GM)^2,\\ \Gamma(\theta)&=\frac{1+\cos^2\theta}{2},\\ \Lambda(\theta)&=\frac{2\sin\theta}{1+\cos^2\theta}. \end{aligned}$$

The NHEK metric in its Poincaré patch is

$$\begin{aligned} ds^2=L^2\Gamma(\theta)\Bigg[& -R^2dT^2+\frac{dR^2}{R^2}+d\theta^2\\ &+\Lambda(\theta)^2(d\Phi+R\,dT)^2 \Bigg]. \end{aligned}$$

The patch $R\gt0$ describes one side of the degenerate horizon at $R=0$. Although the $\mathrm{AdS}_2$-like base is visible, the last term twists the azimuthal circle over it; NHEK is not a direct product.

Enhanced symmetry

Stationarity and axisymmetry give $H_- = \partial_T$ and $m=\partial_\Phi$. Two further vector fields are

$$H_0=T\partial_T-R\partial_R,$$

and

$$\begin{aligned} H_+={}&\left(T^2+\frac{1}{R^2}\right)\partial_T\\ &-2TR\partial_R-\frac{2}{R}\partial_\Phi. \end{aligned}$$

They satisfy

$$\begin{aligned} [H_0,H_-]&=-H_-,\\ [H_0,H_+]&=H_+,\\ [H_-,H_+]&=2H_0, \end{aligned}$$

while $m$ commutes with all three. These are the $\mathfrak{sl}(2,\mathbb R)\oplus\mathfrak u(1)$ commutators. The $R^{-1}\partial_\Phi$ term in $H_+$ is essential: the azimuthal fibration must transform along with the $\mathrm{AdS}_2$ base. This enhanced group was one of the central results of Bardeen and Horowitz 1999, §2, pp. 4–5.

The dilation generated by $-H_0$ is precisely the universal near-horizon scaling. The third generator is not inherited as an independent Kerr isometry; it appears only after the asymptotic region has decoupled.

A horizon-regular patch and its data

Set

$$\begin{aligned} v&=T-\frac1R,\\ \varphi&=\Phi-\ln R. \end{aligned}$$

Then the apparent singularities at $R=0$ combine to give

$$\begin{aligned} ds^2=L^2\Gamma(\theta)\Bigg[& -R^2dv^2+2\,dv\,dR+d\theta^2\\ &+\Lambda(\theta)^2(d\varphi+R\,dv)^2 \Bigg]. \end{aligned}$$

The norm of $\partial_v$ has a quadratic zero, so the horizon is degenerate. Its spatial cross-section has metric

$$\gamma_H=L^2\Gamma(\theta) \left[d\theta^2+\Lambda(\theta)^2d\varphi^2\right].$$

The nonconstant warp factor, the unit fibration coefficient in $d\varphi+Rdv$, and this distorted two-sphere are the characteristic NHEK horizon data. Near either axis, $\Gamma\to1$ and $\Lambda\sim\sin\theta$, so a $2\pi$-periodic $\varphi$ gives no conical defect.

There are two independent vacuum checks. First, every $\lambda\gt0$ metric in the scaling family is diffeomorphic to vacuum Kerr, and the smooth limit must satisfy $R_{ab}=0$. Second, direct substitution into the Ricci tensor gives zero; the cancellation requires the precise functions $\Gamma$ and $\Lambda$, not merely an arbitrary warped $\mathrm{AdS}_2$ ansatz.

Global NHEK is larger than the Poincaré patch

Coordinates analogous to global $\mathrm{AdS}_2$ coordinates put the metric in the form

$$\begin{aligned} ds^2=L^2\Gamma(\theta)\Bigg[& -(1+y^2)d\tau^2+\frac{dy^2}{1+y^2}+d\theta^2\\ &+\Lambda(\theta)^2(d\phi+y\,d\tau)^2 \Bigg]. \end{aligned}$$

With $\tau,y\in\mathbb R$, this chart covers the complete throat and has two timelike conformal boundaries at $y\to\pm\infty$. Constant-$\tau$ surfaces are spacelike, but $\partial_\tau$ is not timelike everywhere because of the rotational fibration. The coordinate transformation, completeness argument, and boundary analysis are in Bardeen and Horowitz 1999, §2, pp. 4–6, Eqs. (2.7)–(2.10).

The global extension contains points not represented by holding a point of the parent Kerr exterior fixed as $\lambda\to0$. It should therefore be used as an exact throat background with its own boundary conditions, not silently identified with the full extremal Kerr spacetime.

What survives the decoupling

The parameter $J$ remains as the overall length scale $L^2=2GJ$; after that factor is removed, the dimensionless NHEK shape is universal. The horizon cross-section, its axial Killing field, and the associated angular momentum survive. By contrast, the parent mass and the value of the angular velocity relative to inertial observers at infinity are not intrinsic NHEK data. The limit has removed those observers and allows constant rescalings of its throat time. This general loss of asymptotic mass and angular-velocity data in a near-horizon limit is explained in Kunduri and Lucietti 2013, §2.4, pp. 15–16.

For extremal vacuum Kerr one can of course infer $M=\sqrt{J/G}$ after adding the knowledge that the throat came from that family. That inference uses the global Kerr relation; it is not a measurement made entirely inside NHEK. The same distinction matters in classification: proving that horizon data are locally NHEK does not by itself construct an asymptotically flat Kerr exterior.

Rotation also prevents a globally timelike Killing field of the most obvious form. In the Poincaré chart,

$$g(\partial_T,\partial_T) =L^2\Gamma R^2\left(\Lambda^2-1\right),$$

which changes sign with latitude. In the global chart, $\partial_\tau$ likewise becomes spacelike in part of the asymptotic throat. The spacetime still has a global time function, but a conserved energy built from one stationary Killing field need not be positive everywhere. This is a geometric warning, not a verdict about stability: a stability statement must specify the field, norm, boundary conditions at the timelike throat boundaries, and whether the background is NHEK or the full Kerr exterior.

Finally, the enhanced $\mathrm{SL}(2,\mathbb R)$ symmetry belongs to the exact limit. At small but nonzero $\lambda$, transverse corrections reconnect the throat to the Kerr exterior and break that symmetry. Results that rely on the full group therefore need a separate matching argument before being applied to finite-radius observations in Kerr.

Solved diagnostic: area and regularity

Compute the area of a horizon cross-section and compare it with extremal Kerr. Then check the north pole.

Solution

The area density is

$$\sqrt{\det\gamma_H}=L^2\Gamma\Lambda=L^2\sin\theta.$$

Therefore

$$\begin{aligned} A_H &=\int_0^{2\pi}d\varphi\int_0^\pi d\theta\,L^2\sin\theta\\ &=4\pi L^2=8\pi GJ, \end{aligned}$$

exactly the extremal Kerr area. Near $\theta=0$, $\Gamma=1+O(\theta^2)$ and $\Lambda=\theta+O(\theta^3)$, so

$$\begin{aligned} \gamma_H=L^2\Big[&d\theta^2+\theta^2d\varphi^2\\ &+O(\theta^2)d\theta^2+O(\theta^4)d\varphi^2\Big]. \end{aligned}$$

The pole is locally a smooth plane in polar coordinates when $\varphi\sim\varphi+2\pi$. The throat preserves the horizon area and axial regularity even though it loses asymptotic flatness.

The near-horizon classification theorems explain under which assumptions NHEK is forced, while perturbations and their boundary conditions remain in Black-Hole Perturbation Theory.

References

• Bardeen, James M., and Gary T. Horowitz. “The Extreme Kerr Throat Geometry: A Vacuum Analog of $\mathrm{AdS}_2\times S^2$.” Physical Review D 60, 104030 (1999). DOI. Open PDF.
• Kunduri, Hari K., and James Lucietti. “Classification of Near-Horizon Geometries of Extremal Black Holes.” Living Reviews in Relativity 16, 8 (2013). DOI.