Nonlinear Instability Mechanisms

A nonlinear instability is a failure of stability for the full evolution equations. Its onset may already be visible as a linearly growing mode, or it may require interactions absent from the linearized equation. This page distinguishes those two possibilities and then asks what backreaction does to the endpoint. Resonant transfer, shocks, and finite-amplitude collapse illustrate different mechanisms. As of the evidence cutoff 2026-08-12, the specified small-data vacuum class near every fixed subextremal Kerr exterior is covered by a nonlinear stability preprint; claims outside that class concern different questions.

Required background. PDEs and Green Functions for Gravitation supplies nonlinear energy transfer and instability criteria. Helpful background. Black-Hole Geometry and Mechanics supplies Kerr horizons; Black-Hole Perturbation Theory supplies linear modes; Formulations of General Relativity supplies nonlinear evolution; Computational Relativity supplies convergence tests.

Instability mechanisms and their domains

Mechanism	Linear seed	Genuinely nonlinear step	Controlled regime
massive-field superradiance	exponentially growing bound mode	backreaction, saturation, collapse, or hair selection	matter model and coupling dependent
mass inflation	inner-horizon blueshift	counterstreaming stress changes mass and curvature	rigorous in spherical charged models; partial near-Kerr control
resonant cascade	neutral or weakly damped spectrum	mode products transfer energy secularly	especially sensitive to reflecting boundaries
shock formation	compressive characteristic field	gradient steepening and caustic interaction	matter and symmetry dependent
critical collapse	one unstable perturbative direction	finite tuning approaches a nonlinear attractor	data-family and symmetry dependent
parameter migration	flux changes mass or spin	trajectory crosses a stability boundary	requires such a boundary in the model

The full-subextremal vacuum result rules out an exterior instability of sufficiently small data in its exact class: smooth constraint-satisfying perturbations, high weighted regularity, specified asymptotic expansion, and $|\mathbf a_0|\lt m_0$ in source units $G=c=1$ relax to a nearby Kerr exterior Hintz 2026, Theorem 1.1, pp. 2–3, and Theorem 13.1, pp. 318–323, PDF. Restoring dimensions, $|\mathbf a_0|/m_0=c|\mathbf J_0|/(GM_{0,{\rm phys}}^2)$. The theorem does not address extremal Kerr, massive bosonic matter, reflecting boundaries, large data, or the complete interior.

Linear growth versus nonlinear selection

For an amplitude vector $A_i(t)$, weakly nonlinear dynamics often has

$$\dot A_i=\lambda_i A_i +\sum_{jk}C_{ijk}A_jA_k +\sum_{jkl}D_{ijkl}A_jA_kA_l+\cdots .$$

If $\operatorname{Re}\lambda_i\gt 0$, the origin is linearly unstable. Nonlinearity determines saturation but not the existence of the instability. Massive-field superradiance is of this type: the bound state grows already in the linearized matter equation Brito, Cardoso, and Pani 2020, §§4.4–4.7 and 5.2–5.6, pp. 50–68 and 89–128, PDF. Calling the whole phenomenon “a nonlinear instability of vacuum Kerr” would be incorrect.

If the homogeneous linear evolution is bounded but nonlinear resonant terms accumulate, the onset can require nonlinearity. Merely knowing $\operatorname{Re}\lambda_i\le0$ is not enough: a linear Jordan block can already produce polynomial growth. Nor does a secular term by itself prove instability; it can resum into a small frequency shift. A quadratic resonance must satisfy frequency and symmetry conditions such as

$$\omega_i=\omega_j+\omega_k, \qquad m_i=m_j+m_k.$$

The coupling coefficient must also be nonzero, and the result depends on damping, phase evolution, and boundary leakage. In asymptotically flat Kerr, radiation and horizon absorption remove energy; in anti-de Sitter space, a reflecting timelike boundary can repeatedly return it. Bizoń and Rostworowski give numerical evidence for a weakly turbulent cascade in the spherical Einstein–scalar AdS model Bizoń and Rostworowski 2011, Eqs. (1)–(9) and pp. 1–4, PDF. That is evidence about AdS boundary conditions, not a counterexample to asymptotically flat Kerr stability.

Backreaction and region dependence

Mass inflation illustrates a different nonlinear mechanism. Test radiation is blueshifted at the Cauchy horizon, but the Hawking-mass divergence arises when counterstreaming stress backreacts on the geometry. Rigorous spherical results control this feedback under a charged matter model Dafermos 2005, §§9–12, pp. 33–51, PDF. The effect concerns the interior, whereas the 2026 Kerr theorem controls an exterior region. The two statements can both be true.

Finite-amplitude collapse and shock formation also lie outside linear orbital stability. A solution may be linearly stable near an equilibrium but possess a distant basin boundary. Small-data stability proves an open stable neighborhood; it does not classify all large data.

What would establish a new mechanism

A serious claim needs:

• a continuum model, asymptotics, matter content, symmetry class, and data topology;
• evidence that the effect is not a gauge mode or drift along a stationary family;
• a comparison with the exact linear spectrum showing whether growth exists already at first order;
• an invariant nonlinear diagnostic and closed energy or charge budget;
• for computation, constraint monitoring, at least three effective resolutions or an equivalent tolerance sequence, and an uncertainty estimate;
• a precise comparison with applicable stability theorems; an apparent contradiction in the same data class requires rechecking both the evolution and the theorem hypotheses.

No computation was rerun for this page. Published numerical evidence is used only within its declared equations, boundary conditions, symmetry, and convergence tests.

The general question remains unresolved beyond the small-data vacuum exterior class. A theorem with broader hypotheses, a stable counterexample on an open data set, or an independently converged nonlinear calculation could clarify those regimes. Corrections to the cited results could also change the conclusions.

Formulation-choice diagnostic

A massive scalar bound state on Kerr grows as $e^{\Gamma t}$ in the linear Klein–Gordon equation, and a nonlinear simulation later settles to a hairy state. Is the instability genuinely nonlinear?

Solution

Its onset is already linear because $\Gamma\gt0$ in the linearized matter problem. If the full coupled evolution leaves a fixed neighborhood of the Kerr solution, that is a nonlinear instability too, but one triggered by a linear mode. Nonlinear backreaction selects or modifies the endpoint—hair, spin-down, radiation, or collapse—rather than creating the initial growing direction. It does not contradict vacuum Kerr stability because the model contains a massive field.

References

• Bizoń, Piotr, and Andrzej Rostworowski, “On Weakly Turbulent Instability of Anti-de Sitter Space”, Physical Review Letters 107, 031102, 2011. Open PDF.
• Brito, Richard, Vitor Cardoso, and Paolo Pani, Superradiance, Lecture Notes in Physics 971, second edition, Springer, 2020. Open PDF.
• Dafermos, Mihalis, “The Interior of Charged Black Holes and the Problem of Uniqueness in General Relativity”, Communications on Pure and Applied Mathematics 58, 445–504, 2005. Open PDF.
• Hintz, Peter, “Nonlinear Stability of Subextremal Kerr Black Holes”, arXiv:2606.28253v2, 2026. Open PDF.