Answers are grounded in this paper and cite their sources.
Theorem. The singly spinning Myers–Perry black hole in seven spacetime dimensions with rotation parameter a/r₊ = 8 (r₊ = 1, a = 8, μ = 65) is linearly unstable in the sense of Prabhu and Wald (arXiv:1501.02522). There are smooth, compactly supported, constraint-satisfying initial data, invariant under U(1) × SO(4) and odd under the (t-φ) reflection, whose gauge-fixed representative P′ has negative potential energy 𝒰(𝒦P′, 𝒦P′). Because the kinetic energy is positive definite, the Rayleigh–Ritz quotient ω² is negative, and the solution's kinetic energy grows faster than e2αt for every α < |ω|. Nothing is claimed for other dimensions or rotations, for growth rates, or for the nonlinear problem. Method. The trial data are built from four potential functions; the two nontrivial linearized momentum constraints are solved pointwise, and the other constraints vanish identically for odd data. The potential energy of the gauge-fixed data is shown to equal, up to a positive factor, the integral of an explicit local quadratic density, using only identities of the Prabhu–Wald framework. That integral is certified negative by adaptive Gauss–Legendre quadrature over 7088 panels with rigorous remainders; the integrand is bounded on complex polydiscs by bivariate Taylor models with interval remainders. A mollification argument passes to smooth data. Margin. The certified value is a small residue of large contributions of both signs (about 0.2 % of either signed contribution); the enclosure is rigorous, but the sign relies on the exactness of the symbolic derivation of the integrand. The article states this dependence and the tests that support it. Trust assumptions. FLINT (Arb ball arithmetic and exact rational arithmetic) as used through python-flint 0.9.0, IEEE binary64 and standard matrix-product error bounds; the SymPy symbolic derivation; the results of Prabhu and Wald as published; standard analysis and causality theory; and the provenance of the archived certificate records (removable by the full rerun in the supplement). Files. The article (PDF) and a supplementary archive with: the certificate programs, their inputs and the records of all panels; the complete working proof and its earlier versions, the symbolic derivation and exact checks; a one-step check (check.py; --replay replays the certificate, --replay --full reruns it); the reviews and independent checks; the trust list. AI tools. AI systems (Claude Opus 5.5, Claude Fable 5.2 and GPT-6 Astra) did the computational and drafting work under the author's direction; the article's section on AI use describes it. Blind reviews crossed developers: what one developer's model drafted, the other developer's model examined. Models of both families re-proved the theorem starting from its statement. Two adversarial reviews and a recomputation with a second interval library were carried out. No human expert has yet checked the proofs line by line.
No takes yet. Share an insight, caveat, or question.
Neko2333OWO (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: