Turbulent Instability of Anti-de Sitter

The AdS instability problem asks whether arbitrarily small perturbations can accumulate nonlinear interactions until a trapped surface forms. The classic spherical scalar calculation supplies a compelling cascade mechanism, but “AdS is unstable” is not one theorem with model-independent quantifiers. Matter content, symmetry, boundary conditions, data topology, phase coherence, and the meaning of genericity all change the statement.

Helpful background. Black-Hole Geometry and Mechanics supplies trapped-surface criteria; Black-Hole Perturbation Theory supplies spectral language; Formulations of General Relativity supplies the constrained evolution; Exact Solutions of Einstein’s Equations supplies the AdS background; Causality and Global Structure supplies the timelike boundary; and Computational Relativity supplies convergence standards for delayed collapse.

The weakly turbulent mechanism

Consider four-dimensional Einstein gravity with $\Lambda=-3/L^2$, a real massless scalar field, spherical symmetry, regular center, and reflecting conformal boundary. For small initial amplitude $\varepsilon$, expand

$$\phi=\varepsilon\phi_1+\varepsilon^3\phi_3+\cdots, \qquad g=g_{\mathrm{AdS}}+\varepsilon^2h_2+\cdots .$$

The linear field is a sum of normal modes with evenly spaced frequencies $\omega_jL=3+2j$. Metric backreaction first appears at order $\varepsilon^2$; at order $\varepsilon^3$, products of modes can drive another mode resonantly. With $t$ a physical time, a secular term of size $\varepsilon^3(t/L)$ becomes comparable to the leading $O(\varepsilon)$ field at

$$t\sim L\varepsilon^{-2}.$$

Introducing slow time $\tau=\varepsilon^2t/L$ resums these terms into coupled amplitude equations. Resonant quartets obey $\omega_i+\omega_j=\omega_k+\omega_l$, but a nonzero coupling coefficient and coherent phases are also needed. A frequency shift alone can remove some secular terms without causing a cascade. Transfer toward large $j$ puts energy into shorter wavelengths and can increase local gradients even while the total mass stays constant. “Weakly turbulent” refers to this spectral transfer, not a Kolmogorov fluid spectrum. Once strong focusing invalidates the small-amplitude expansion, trapped-surface formation must be checked in the full evolution.

Bizoń and Rostworowski evolved localized Gaussian data and found delayed apparent-horizon formation after increasing numbers of boundary reflections, an onset time proportional to $\varepsilon^{-2}$, and transfer toward high frequencies Bizoń and Rostworowski 2011, Eq. (9) and Figs. 1–3, pp. 2–4. These are numerical observations in one constrained model. Even there, finite runtime cannot prove collapse for every nonzero amplitude.

Evidence does not support one universal quantifier

Evidence cutoff: 12 August 2026. The main positions can be compared without forcing them into a false yes-or-no verdict.

Formulation or claim	Positive evidence	Boundary of the claim
Generic localized spherical scalar data collapse at arbitrarily small amplitude	Long-time evolutions show repeated focusing, a forward mode cascade, and increasingly delayed horizon formation	The simulations sample finite-dimensional families at finite resolution and time; stability islands obstruct “all data”
Small periodic configurations and long-lived neighborhoods can avoid the cascade	Perturbative and numerical time-periodic constructions, with direct evolution tests, support regular alternatives	These calculations are not a proof of global regularity of whole neighborhoods, their measure, or their topology
Reflecting AdS is nonlinearly unstable for a well-posed Einstein–matter system	A rigorous construction gives arbitrarily small spherically symmetric Einstein–massless-Vlasov data whose developments contain trapped spheres	The theorem is existential, uses Vlasov matter and a particular scale-invariant norm, and is not a generic scalar-field or vacuum theorem
Exact resonance is the decisive cause	The evenly spaced scalar spectrum explains secular interactions and the slow timescale	Phase coherence, focusing, boundary flux, and nonlinear selection rules also matter; modifying the spectrum does not produce one universal outcome
Vacuum AdS is generically unstable	Nonspherical vacuum perturbation theory exhibits third-order secular resonances and direct and inverse cascades Dias and Santos 2018, §§I and III.C, pp. 1–3, 11–14	This perturbative evidence does not classify generic nonlinear data; spherical vacuum gravity has no radiative degree of freedom

The stable side deserves equal scrutiny. Maliborski and Rostworowski construct time-periodic Einstein–scalar configurations perturbatively and numerically, then test them by direct evolution Maliborski and Rostworowski 2013, pp. 2–5, PDF. Martinon reviews these constructions and vacuum geon analogues in Martinon 2017, §§4.2–4.6, pp. 18–29. This is substantial evidence against universal collapse, but numerical periodicity and long survival must not be relabeled a theorem of global nonlinear stability.

The rigorous instability result is equally important and equally specific. Moschidis proves that there exists a one-parameter family of smooth, spherically symmetric, asymptotically AdS initial data for the Einstein–massless Vlasov system, converging to AdS in a declared scale-invariant norm, such that every nontrivial member forms a trapped sphere Moschidis 2023, §1.1, Theorem 1, p. 5. This establishes nonlinear instability for that matter system and topology. It neither says that every small datum collapses nor substitutes for a theorem about the Einstein–scalar or vacuum equations.

What a decisive classification would require

A useful statement must define five ingredients:

1. Evolution problem: vacuum, scalar, Vlasov, or another matter model, with dimension and symmetry class.
2. Boundary problem: reflecting, dissipative, mixed, or driven data, including the gravitational boundary degrees of freedom.
3. Smallness topology: the norm in which data approach AdS and the regularity needed for well-posedness.
4. Outcome: loss of a norm, curvature blow-up, trapped-surface formation, or failure of global regularity. These are not interchangeable.
5. Quantifier: existence, an open set, a dense set, full measure, or all sufficiently small data.

The open question is therefore not whether any instability mechanism exists—it does—but how stable and collapsing sets are arranged for each formulation, especially for vacuum gravity without symmetry. A resolution would need either a theorem classifying an open or generic set in a stated topology, or numerical evidence that controls the joint limits of amplitude, resolution, and evolution time while mapping stable islands rather than missing them.

An update should be triggered by a new rigorous genericity theorem, a counterexample invalidating an existing formulation, or a validated long-time survey that materially changes the inferred topology of stable and collapsing sets. A longer evolution of one Gaussian family alone would refine evidence, not settle the general conjecture.

Exercise: repair an instability claim

A paper states, “Every sufficiently small perturbation of AdS eventually forms a black hole.” Identify at least four missing specifications, then give one established result that is strong enough to cite and narrow enough to be correct.

Solution

The claim must specify the field equations and matter, spacetime dimension, symmetry class, conformal-boundary condition, initial-data regularity and norm, meaning of “sufficiently small,” definition of black-hole formation, and quantifier over data. It must also distinguish mathematical time from the finite runtime of a simulation.

One correct replacement is: in four-dimensional spherical Einstein–massless Vlasov theory with reflecting AdS boundary conditions, Moschidis constructs a family of smooth data approaching AdS in a particular scale-invariant norm such that every nonzero family member develops a trapped sphere. This is an existential nonlinear-instability theorem. It is not a statement about every small datum, a scalar field, or vacuum perturbations.

References

• Bizoń, Piotr, and Andrzej Rostworowski. “On Weakly Turbulent Instability of Anti-de Sitter Space.” Physical Review Letters 107 (2011): 031102. doi:10.1103/PhysRevLett.107.031102. Open PDF.
• Dias, Oscar J. C., and Jorge E. Santos. “AdS Nonlinear Instability: Breaking Spherical and Axial Symmetries.” Classical and Quantum Gravity 35 (2018): 185005. doi:10.1088/1361-6382/aad514. Open PDF.
• Maliborski, Maciej, and Andrzej Rostworowski. “Time-Periodic Solutions in an Einstein AdS–Massless-Scalar-Field System.” Physical Review Letters 111 (2013): 051102. doi:10.1103/PhysRevLett.111.051102. Open PDF.
• Martinon, Grégoire. “The Instability of Anti-de Sitter Space-Time.” arXiv:1708.05600v3 (2017), 58 pp. Abstract. Open PDF.
• Moschidis, Georgios. “A Proof of the Instability of AdS for the Einstein–Massless Vlasov System.” Inventiones Mathematicae 231 (2023): 467–672. doi:10.1007/s00222-022-01152-7. Open PDF.