Symmetries of extremal horizons

We prove an intrinsic analogue of Hawking’s rigidity theorem for extremal horizons in arbitrary dimensions: any compact cross-section of a rotating extremal horizon in a spacetime satisfying the null energy condition must admit a Killing vector field. If the dominant energy condition is satisfied for null vectors, it follows that an extension of the near-horizon geometry admits an enhanced isometry group containing $SO(2,1)$ or the 2D Poincaré group $\mathbb{R}^2 \rtimes SO(1,1)$. In the latter case, the associated Aretakis instability for a massless scalar field is shifted by one order in the derivatives of the field transverse to the horizon. We consider a broad class of examples including Einstein-Maxwell(-Chern-Simons) theory and Yang-Mills theory coupled to charged matter. In these examples we show that the symmetries are inherited by the matter fields.

💡 Research Summary

The paper establishes an intrinsic analogue of Hawking’s rigidity theorem for extremal horizons that holds in arbitrary spacetime dimensions and for a broad class of matter models. Starting from an extremal Killing horizon $H$, the authors focus on a compact, boundary‑less cross‑section $M$ and introduce the intrinsic horizon data: a Riemannian metric $g$, a one‑form $X$, a symmetric tensor $T$ (the pull‑back of the spacetime energy‑momentum tensor), and a scalar $U$ defined by $T(k,\cdot)=Uk$ on the horizon. These quantities satisfy the “horizon equations” (1.1), which are simply the Einstein equations projected onto the horizon. The horizon is called rotating when $X$ is not exact; otherwise it is static.

Two energy conditions are imposed only on null vectors: (EC1) $T(\ell,\ell)\ge0$ (null energy condition) and (EC2) $T(\ell,\cdot)$ is causal or zero (a stronger condition). Lemma 3 shows that if the ambient spacetime satisfies EC1 or EC2, then the associated near‑horizon geometry—constructed from $(g,X,T,U)$ as a spacetime on $\mathbb R^2\times M$ with metric (2.11a)—inherits the same condition.

The first main result (Theorem 1) proves that EC1 alone forces the existence of a Killing vector field $K$ on the compact cross‑section $M$. This is an intrinsic “rotating implies axisymmetric” statement that does not rely on global stationarity or analyticity, unlike the classic Hawking rigidity theorem. The second result (Theorem 2) adds EC2 and shows that $K$ preserves the full horizon data $(X,T,U)$ and extends to a Killing field of the full near‑horizon geometry. Moreover, the near‑horizon geometry acquires an enhanced isometry group containing the orientation‑preserving isometries of either $AdS_2$, Minkowski $\mathbb R^{1,1}$, or $dS_2$. If the horizon is rotating and EC2 holds, only the $AdS_2$ case is allowed, and the isometry group acquires an extra $U(1)$ factor coming from $K$.

A crucial scalar $A$ appears in the analysis; it is shown to be constant on $M$ (Proposition 5) and identified with the Gaussian curvature of the two‑dimensional factor in the near‑horizon metric. Depending on the sign of $A$, the 2‑D factor is $AdS_2$ ($A>0$), flat Minkowski ($A=0$), or $dS_2$ ($A<0$). The constant $A$ plays the role of an extremal analogue of surface gravity: it vanishes for “doubly degenerate” horizons (e.g., the ultracold Reissner‑Nordström‑de Sitter solution) and is non‑zero otherwise. This links directly to the Aretakis instability: for $A\neq0$ the first transverse derivative of a massless scalar field is conserved on the horizon and higher derivatives grow polynomially in the affine parameter $v$. When $A=0$, Proposition 10 shows that the instability is shifted by one order: certain combinations of the field and its first two transverse derivatives are conserved, while a third derivative generically blows up, provided the field itself decays. This “one‑order shift” is interpreted as a signature of the doubly degenerate nature of the horizon.

The authors then examine a wide range of matter theories—Einstein–Maxwell (with possible Chern–Simons term), supergravity reductions, Yang–Mills, and charged scalar or fermion fields. They verify that the matter equations imply the constraints (2.6) and (2.8) on the horizon data, and that the energy conditions (EC1)–(EC2) are satisfied for null vectors. Consequently, the Killing vector $K$ constructed in Theorem 1 also preserves the matter fields, and the full set of fields (metric, gauge potentials, scalar profiles) is invariant under the enlarged isometry group of the extended near‑horizon geometry (Theorem 2). This demonstrates that the intrinsic rigidity and symmetry enhancement are robust under the inclusion of realistic matter content.

Finally, the paper provides explicit calculations of $A$ for the extremal Kerr–Newman–de Sitter family (Appendix A), confirming that $A$ reduces to the usual surface gravity in the non‑extremal limit and vanishes precisely for the doubly degenerate configurations. The authors argue that $A$ should be viewed as the extremal counterpart of the zeroth law of black‑hole mechanics, and it also appears in a Smarr‑type relation for near‑horizon geometries.

In summary, the work delivers a comprehensive, dimension‑independent framework showing that compact rotating extremal horizons necessarily possess a Killing symmetry, that this symmetry extends to an enlarged isometry group of the near‑horizon geometry under mild energy conditions, and that the associated scalar $A$ governs both the geometric classification (AdS$_2$, Minkowski, dS$_2$) and the behavior of Aretakis‑type instabilities. The results hold for a broad spectrum of matter models, making them highly relevant for studies of extremal black holes, supergravity solutions, and the dynamics of fields on such backgrounds.