From Minkowski Stability to Kerr Stability

Minkowski and Kerr stability share a small-data quasilinear-wave strategy, but Kerr adds four structural difficulties: an event horizon, trapped null geodesics, an ergoregion with superradiant energy exchange, and a nontrivial family of stationary endpoints. These are not small lower-order corrections. They change the energy, the decay hierarchy, the gauge construction, and even the statement of what convergence means.

Required background. PDEs and Green Functions for Gravitation supplies hyperbolic estimates and decay. Helpful background. Black-Hole Geometry and Mechanics supplies horizons and trapping; Black-Hole Perturbation Theory supplies linear Kerr dynamics; Formulations of General Relativity supplies gauge reduction; Computational Relativity supplies nonlinear comparison evidence.

Flat-space dispersal

For small asymptotically flat vacuum data near Euclidean data on $\mathbb R^3$, Minkowski stability proves global geodesic completeness and decay toward flat spacetime after a suitable coordinate choice. The perturbation obeys a quasilinear wave system whose nonlinearities possess enough null structure that weak decay remains integrable. There is no event horizon, no black-hole parameter to modulate, and no compact trapped set.

With time measured in a fixed reference length, the central energy schematic is

$$E_N(t)+\int_0^t\mathcal D_N(s)\,ds \le E_N(0) +\int_0^t\mathcal N_N(s)\,ds,$$

where $\mathcal D_N$ is an integrated decay term and $\mathcal N_N$ contains nonlinear errors. Weighted vector-field estimates make $\mathcal N_N$ integrable for sufficiently small data. Christodoulou and Klainerman organize the geometry through null foliations and exploit the structure of the Einstein equations to control these nonlinear interactions Christodoulou and Klainerman 1993, Chapters 1–3.

Horizon, trapping, and rotation

On subextremal Kerr, energy can leave through two boundaries: future null infinity and the event horizon. The horizon is helpful because its redshift supplies a positive nondegenerate estimate transverse to it. That estimate degenerates toward extremality.

Trapping is harmful. Null geodesics can remain near the photon region, so integrated local energy decay degenerates there or loses derivatives. On rotating Kerr the trapped set is not a single sphere independent of frequency. A nonlinear proof must absorb this loss through stronger microlocal control, commutation, or a hierarchy with additional regularity.

Rotation also creates an ergoregion where the stationary Killing field is spacelike. Its associated energy need not be positive, and individual modes satisfying the superradiant condition can extract rotational energy. Mode stability rules out growing vacuum gravitational separated modes with the prescribed horizon and infinity conditions; quantitative energy and decay estimates are additional requirements. Klainerman and Szeftel summarize these obstacles in Klainerman and Szeftel 2022, §§2.4–2.6, pp. 10–15, PDF.

Parameters and zero-frequency structure

Minkowski space has asymptotic Poincaré symmetries, and small vacuum perturbations approach flat geometry locally as radiation disperses. This does not mean that their ADM mass is zero: conserved energy can move to larger radii, so a local late-time limit and a spatial-infinity limit need not commute. Kerr instead leaves a stationary black hole with nonzero remnant charges. Linearized perturbations include:

$$\delta M\,\partial_M g_{M,a}, \qquad \delta a^i\,\partial_{a^i}g_{M,\mathbf a}, \qquad \mathcal L_X g_{M,\mathbf a}.$$

The first two are physical parameter variations; the last is pure gauge. None should be forced to decay as radiation. At nonlinear order, flux changes the asymptotic mass and spin, while boosts and center-of-mass motion complicate comparison at infinity. A proof must project or modulate these modes and solve for the final parameters.

Low frequency is therefore decisive. Hintz’s analysis identifies large zero-energy states of the gauge potential, constraint propagation, and linearized Einstein operators, then augments the system so the nonlinear iteration can determine the needed finite-dimensional corrections Hintz 2026, §§1.3–1.5, pp. 9–42, and §§6–10, pp. 113–217, PDF.

What carries over and what does not

The shared proof skeleton is:

1. solve the constraints and fix a hyperbolic gauge;
2. establish high-order energy boundedness and integrated decay;
3. extract asymptotic parameters and pure-gauge contributions;
4. estimate quadratic and higher errors with a bootstrap or tame iteration;
5. improve the assumed bounds and continue globally.

But one cannot obtain Kerr stability by inserting a small stationary metric error into the Minkowski theorem. The Kerr exterior has different causal boundaries, trapping, long-range tails, nonpositive stationary energy, and nondecaying parameter modes.

For ultimately Schwarzschildean data, Dafermos, Holzegel, Rodnianski, and Taylor give a nonlinear theorem and detailed proof architecture under their stated codimension and asymptotic assumptions Dafermos et al. 2021, Theorem I.3.1, pp. 5–8, and §V, pp. 21–28, PDF. Hintz’s 2026 theorem covers perturbations of every fixed subextremal Kerr background, but still assumes vacuum, smallness, high regularity, and controlled asymptotic expansion Hintz 2026, Theorem 1.1, pp. 2–3, and Theorem 13.1, pp. 318–323, PDF.

Failure boundary

Near extremality, the horizon redshift degenerates and new long-lived behavior appears. With matter, additional bound states or nonlinear couplings may be present. For large data, topology change or new trapped regions can occur. None lies inside the hypotheses of the vacuum small-data theorem.

One must also compare asymptotic data classes, not just spin ranges. Hintz’s theorem allows slowly decaying terms with specified power–logarithm expansions, while some earlier theorems allow less structured remainders with different decay rates. The full-spin theorem therefore does not simply contain every earlier data class Hintz 2026, p. 2, footnote 1, PDF.

Check your understanding

Why can a proof that controls only a positive energy associated with the stationary Kerr Killing field fail, even when all initial perturbations are small?

Solution

Inside the ergoregion the stationary Killing field is spacelike, so its energy density is not pointwise positive. Superradiant modes can carry negative Killing energy into the horizon while positive energy escapes. The proof needs a modified positive current, frequency localization, or related estimates, together with mode stability and redshift control. Small amplitude alone does not repair the sign.

References

• Christodoulou, Demetrios, and Sergiu Klainerman, The Global Nonlinear Stability of the Minkowski Space, Princeton University Press, 1993.
• Dafermos, Mihalis, Gustav Holzegel, Igor Rodnianski, and Martin Taylor, “The Nonlinear Stability of the Schwarzschild Family of Black Holes”, arXiv:2104.08222v1, 2021. Open PDF.
• Hintz, Peter, “Nonlinear Stability of Subextremal Kerr Black Holes”, arXiv:2606.28253v2, 2026. Open PDF.
• Klainerman, Sergiu, and Jérémie Szeftel, “Brief Introduction to the Nonlinear Stability of Kerr”, arXiv:2210.14400v2, 2022. Open PDF.