Near-Horizon Classification Theorems

Near-horizon classification asks which smooth horizon data solve the reduced field equations under specified dimension, matter, symmetry, topology, and compactness assumptions. In four-dimensional vacuum gravity, those hypotheses can force the near-horizon extreme Kerr (NHEK) family. Older rotating classifications assumed axial symmetry; intrinsic-rigidity theorems now derive that symmetry from compact horizon data in four-dimensional vacuum and Einstein–Maxwell theory. Removing compactness, changing dimension or matter, or asking for a complete parent spacetime still changes the problem.

Helpful background. Exact Solutions of Einstein’s Equations supplies the Kerr, Kerr–Newman, and higher-dimensional benchmark families; Causality and Global Structure supplies topology and global-extension language; The Einstein Equation supplies the horizon equations; Black-Hole Perturbation Theory treats stability rather than classification; Black-Hole Geometry and Mechanics supplies degenerate horizons and their near-horizon data.

The equations being classified

In Gaussian-null coordinates $(v,r,x^a)$, write the exact near-horizon metric as

$$ds^2=2\,dv\!\left(dr+r h_a(x)dx^a +\frac12r^2F(x)dv\right)+\gamma_{ab}(x)dx^a dx^b.$$

Here $\gamma_{ab}$ is the Riemannian metric on the $(D-2)$-dimensional horizon cross-section $H$, $h_a$ is a one-form on $H$, and $F$ is a scalar on $H$. For a vacuum near-horizon geometry with cosmological constant $\Lambda$, these data obey

$$R_{ab}=\frac12h_ah_b-\nabla_{(a}h_{b)}+\Lambda\gamma_{ab},$$

with

$$F=\frac12h_ah^a-\frac12\nabla_ah^a+\Lambda.$$

A classification theorem identifies solutions of these equations modulo diffeomorphism and the residual near-horizon gauge freedom. It is not a uniqueness theorem for full stationary exteriors. Matter theories replace the right-hand side with stress-tensor data and add horizon matter equations.

Compactness is mathematically active. It makes boundary terms vanish, permits maximum-principle and integral arguments, and constrains orbit-space endpoints. Smoothness at an axis supplies boundary conditions for the ordinary differential equations produced by rotational symmetry.

Static vacuum theorem

A near-horizon geometry is static when its stationary Killing field is hypersurface orthogonal. For compact $H$, the vacuum result is:

If $\Lambda\le0$, every smooth static vacuum near-horizon geometry has $h=0$, $F=\Lambda$, and $R_{ab}=\Lambda\gamma_{ab}$. In four spacetime dimensions the same conclusion also holds for $\Lambda\gt0$.

This is Kunduri and Lucietti 2013, §4.1, Theorem 4.1, p. 25. The proof writes the staticity conditions locally as $h=d\lambda$ and $dF=Fh$, substitutes them into the horizon equations, and applies compactness to a scalar equation for $e^{-\lambda/2}$. The result is a direct-product near-horizon geometry whose cross-section is Einstein. Without compactness the maximum-principle step is lost, and warped solutions occur.

The conclusion does not say that every compact Einstein manifold produces a global static extremal black hole. It only classifies the allowed local throat data in this hypothesis class.

Static Einstein–Maxwell theorem

A nontrivial Maxwell field changes the static branch even in four dimensions. The vanishing-field cases belong to the vacuum branch just stated. Let a smooth static near-horizon geometry with nonzero Maxwell field solve the four-dimensional Einstein–Maxwell–$\Lambda$ equations and have a compact horizon cross-section $H$. Then:

For $\Lambda\ge0$, the geometry is $\mathrm{AdS}_2\times S^2$. For $\Lambda\lt0$, it is $\mathrm{AdS}_2\times H$, where $H$ is one of the constant-curvature surfaces $S^2$, $T^2$, or $\Sigma_g$.

This is Kunduri and Lucietti 2013, §6.2, Theorem 6.3, p. 40. Here $T^2$ is the compact flat case, while $\Sigma_g$ denotes a compact hyperbolic surface of genus $g\gt1$; the spherical case has positive curvature. At $\Lambda=0$, the extremal Reissner–Nordström throat is the familiar spherical representative. For $\Lambda\lt0$, the toroidal and higher-genus possibilities are a concrete reason not to identify the entire compact static electrovacuum branch with a spherical Reissner–Nordström–AdS family.

As in the vacuum theorem, this statement classifies the near-horizon data, not the existence or uniqueness of a complete exterior realizing each topology. Compactness is again part of the theorem; dropping it admits an additional warped family beyond these direct products.

Symmetry enhancement theorem

Enhanced $\mathrm{AdS}_2$-type symmetry is also conditional. Suppose a $D$-dimensional spacetime contains a degenerate horizon invariant under an orthogonally transitive $\mathbb R\times\mathrm U(1)^N$ action, with $1\le N\le D-3$, and suppose the two-dimensional surfaces orthogonal to the group orbits are simply connected. Then its near-horizon geometry has

$$G\times\mathrm U(1)^N,$$

where $G$ is either $\mathrm{SO}(2,1)$ or the two-dimensional Poincaré group. If the strong energy condition holds and the throat is nonstatic, the Poincaré case is excluded. The complete assumptions and conclusion are Kunduri and Lucietti 2013, §3.3, Theorem 3.6, p. 24.

Orthogonal transitivity and simple connectedness are not decorative. They permit the metric to be reduced to a warped two-dimensional orbit-space form and rule out an obstruction to defining the required potentials globally. The theorem is therefore not a proof that every degenerate horizon in every matter theory automatically has $\mathrm{SO}(2,1)$ symmetry.

Axisymmetric four-dimensional rotating uniqueness

In $D=4$ vacuum Einstein gravity, assume a smooth degenerate horizon is invariant under $\mathbb R\times\mathrm U(1)$, the near-horizon geometry is nonstatic, and its cross-section is compact. Then it is the near-horizon limit of extremal Kerr when $\Lambda=0$, or of extremal Kerr–(anti-)de Sitter for nonzero $\Lambda$. This is the precise content of Kunduri and Lucietti 2013, §4.3, Theorem 4.3, p. 26.

The proof reduces the horizon equations to ordinary differential equations on the one-dimensional orbit space $H/\mathrm U(1)$. Smoothness where the axial Killing field vanishes fixes the endpoint behavior; compactness and the field equations eliminate the remaining branches. For $\Lambda=0$, the resulting nonstatic geometry is NHEK.

Einstein–Maxwell theory has the analogous rotating statement: every smooth, axisymmetric, nonstatic $D=4$ Einstein–Maxwell–$\Lambda$ near-horizon geometry with compact cross-section is the near-horizon geometry of extremal Kerr–Newman–$\Lambda$ Kunduri and Lucietti 2013, §6.2, Theorem 6.4, p. 40. The matter model and its coupling are part of the theorem; changing them can create additional branches.

Intrinsic rigidity removes assumed axisymmetry

The preceding theorems historically began with a $\mathrm U(1)$ symmetry. That is no longer an independent hypothesis for compact four-dimensional vacuum or Einstein–Maxwell horizon data. Dunajski and Lucietti prove that a compact, orientable, boundaryless Riemannian cross-section satisfying the vacuum horizon equation with non-gradient $h$ admits a nonzero Killing field. In two dimensions, this field also preserves $h$ for any $\Lambda$. On a spherical cross-section its flow supplies the axial symmetry required by the older classification, so the nonstatic vacuum solution is the extremal Kerr–(A)dS horizon Dunajski and Lucietti 2026, Theorems 1.1–1.2.

Colling, Katona, and Lucietti establish the Einstein–Maxwell counterpart. Let $(H,\gamma)$ be a compact, boundaryless, oriented two-manifold carrying non-gradient quasi-Einstein–Maxwell data. Their theorem constructs a positive function $\Gamma$ for which

$$K=\Gamma h^\sharp+\nabla\Gamma$$

is Killing and preserves both $h$ and the Maxwell data. The resulting $K$ must have a zero, so a corollary forces $H\cong S^2$ and makes its flow an axial $\mathrm U(1)$ action. This reduces the problem to the already classified axisymmetric branch, yielding the extremal Kerr–Newman–(A)dS horizon Colling, Katona, and Lucietti 2025, Theorems 1.1–1.2. Together with the static vacuum and Einstein–Maxwell branches stated above, this completes the compact four-dimensional vacuum and Einstein–Maxwell near-horizon classification. It does not establish axial symmetry of an entire parent exterior.

Where the classifications stop

Four limitations prevent overreading these results.

Classification and symmetry have different domains. The symmetry-free classification completed above uses compact, orientable, boundaryless two-dimensional cross-sections and the four-dimensional vacuum or Einstein–Maxwell–$\Lambda$ equations. That Kerr or Kerr–Newman classification is not dimension-independent or matter-independent. A broader preprint theorem, revised in August 2026, shows that in arbitrary dimension rotating horizon data on a compact boundaryless cross-section admit a Killing field preserving all the horizon data when the associated near-horizon geometry satisfies the null energy condition. The field therefore extends to the near-horizon spacetime Colling 2025, revised 2026, Theorems 1–2. This symmetry theorem does not classify the resulting geometries and does not turn a local horizon Killing field into a global axial symmetry of an arbitrary exterior.

Noncompact cross-sections. Black branes and planar horizons remain valid near-horizon problems, but the compactness arguments fail. The compact classification cannot be applied by replacing a sphere with a plane after the proof.

Higher dimensions. Even with two rotational symmetries, five-dimensional vacuum gravity at $\Lambda=0$ admits compact cross-sections of topology $S^1\times S^2$, $S^3$, and lens-space quotients, with boosted Kerr-string, Myers–Perry, and Kaluza–Klein families. The exact list and assumptions are Kunduri and Lucietti 2013, §4.4, Theorem 4.5, p. 29. NHEK uniqueness is not dimension-independent.

No guaranteed parent solution. A smooth solution of the horizon equations is necessary data for a smooth extremal black hole, but it may fail to extend to a complete exterior with desired charges, asymptotics, and causal structure. Classification can rule out parent horizons whose data never appear; it cannot generally construct a parent from every allowed throat.

Solved diagnostic: repair an overstrong claim

Consider the statement: “Every smooth nonstatic four-dimensional vacuum degenerate horizon has NHEK near-horizon geometry.” Is it a consequence of the classification theorem? If not, formulate a supported version.

Solution

No. The statement omits a compact, orientable, boundaryless spatial cross-section and does not state $\Lambda=0$, which is needed for NHEK rather than Kerr–(A)dS. A supported version is: “In four-dimensional vacuum Einstein gravity with $\Lambda=0$, every smooth nonstatic near-horizon geometry with a compact, orientable, boundaryless cross-section is isometric to the near-horizon limit of extremal Kerr.” Intrinsic rigidity and the two-dimensional topology restriction supply the axial $\mathrm U(1)$ action; the older rotating classification then determines the geometry. This concludes local throat geometry, not existence or uniqueness of a complete asymptotically flat parent exterior. Without the stated cross-section hypotheses, the integral rigidity argument and the classification do not apply.

The extremal-horizon instability interface next separates these stationary classification results from evolution. Full decay estimates and nonlinear dynamics remain with Black-Hole Perturbation Theory.

References

• Colling, Alex, David Katona, and James Lucietti. “Rigidity of the Extremal Kerr–Newman Horizon.” Letters in Mathematical Physics 115, 19 (2025). DOI.
• Colling, Alex. “Symmetries of Extremal Horizons.” arXiv:2512.10200v2 [gr-qc] (2025; revised 2026), to appear in Communications in Mathematical Physics. arXiv.
• Dunajski, Maciej, and James Lucietti. “Intrinsic Rigidity of Extremal Horizons.” Journal of Differential Geometry 132, no. 1 (2026): 179–201. 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.