{"id":"9fbda826-609f-4736-8c73-a8419d9af96d","arxiv_id":"2411.12568","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Small time-decaying perturbations of stationary black hole spacetimes admit a unique smooth null horizon asymptotic to the unperturbed horizon, equal to the event horizon under global assumptions.","lead":"Mathematicians often want to know whether a black hole that is settling down to a stationary state still has a sharp, smooth horizon. This paper proves that, for a broad class of such asymptotically stationary spacetimes, a unique null hypersurface persists, asymptotic to the unperturbed horizon, and in the Kerr-Newman cases it is the event horizon.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The local horizon construction in Theorem 3.2 is plausible; the load-bearing weak point is Theorem 3.4's event-horizon identification, which depends on an unproved global far-field escape condition, the assumed radius R0.","rationale":"The reader's CONDITIONAL verdict matches my reading. Theorem 3.2 is the main technical result and its dynamical-systems structure is coherent: the normal-sink computation in Lemma 3.1 and the invariant-manifold theorem in Section 2 give a convincing local construction of a unique smooth null hypersurface asymptotic to H0. The global upgrade in Theorem 3.4 is separate: it requires a far-field escape condition, the radius R0, which is assumed rather than derived. That assumption is genuinely load-bearing for the abstract's Kerr-Newman conclusion, because without it the inclusion E subset J^-(I+) can fail and H need not be the boundary of the black hole region. This is a limitation, not an internal inconsistency. I also examined whether the local proof of Theorem 3.2 contains a fatal gap. The only suspicious step is the transport argument in the nullness reduction, where the equality g(L,X) = 0 at t* = t1 is asserted for all X in TH without proof; this is not a general property of hypersurfaces ruled by null geodesics. However, the subsequent decomposition argument M cap {t* >= t1} = B disjoint H disjoint E gives an independent proof that T_z H cannot be timelike, so the theorem survives without relying on the questionable step. My stress test therefore does not move the reader's conditional verdict.","tokens_in":12712,"tokens_out":41743,"duration_ms":463373,"concrete_test":"Take g0 to be Schwarzschild in the coordinates of Theorem 3.4 and add a compactly supported, t-dependent perturbation h_tt = A phi((r-R1)/w)(1+t)^(-alpha) near a large radius R1, with A large enough that at t = 0 the shell is strongly focusing for radial null geodesics. Numerically check whether some null geodesic starting at a point with r > R1, or from inside R1 after crossing the shell, fails to reach I+, while g - g0 remains in rho C^infinity_b with rho = (1+t)^(-alpha). If such a spacetime exists, R0 is an independent assumption and Theorem 3.4's conclusion can fail; if none exists, the assumption may be redundant.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract's strongest physical claim, that in the Kerr(-Newman) case the constructed H is the event horizon, rests on Theorem 3.4. Its final step uses the assumption that every point in {t* >= 0, r >= R0} can be connected to future null infinity I+ by a future causal curve. This condition is not derived from the local decay hypothesis g - g0 in rho C^infinity_b nor from the normal-sink dynamics; it is an additional global restriction on the far field. In the proof, R0 enters only to show E subset J^-(I+) cap {t* >= t0}. Without it, the inclusions B cap J^-(I+) = empty and H cap J^-(I+) = empty still hold, but E may contain far-field trapped points, so H need not be the boundary of J^-(I+). Remark 3.5 gives sufficient conditions for R0, but it does not prove they follow from the theorem's hypotheses, nor that R0 is necessary. Thus the advertised event-horizon conclusion is conditional on a global escape property lying outside the local dynamical theorem. A separate, secondary issue is the nullness-reduction step asserting g(L,X) = 0 at t* = t1 for all X in TH without justification; for a generic Lorentzian hypersurface ruled by null geodesics this is false. The later B/E decomposition argument provides an independent route to nullness, so this does not affect Theorem 3.2 itself.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper develops a general unstable-manifold theorem (Theorem 2.2) for smooth maps that translate in time and have a normal sink at an invariant manifold, extending the machinery of [Hin21] and [HPS77]. It then applies this theorem to the time-1 null geodesic flow lifted to the spherical cotangent bundle near the conormal bundle of a non-degenerate horizon in Kerr-type spacetimes. Theorem 3.2 establishes that for any Lorentzian metric g on a neighborhood of the horizon with g − g0 ∈ ρC∞b, there exists a unique smooth null hypersurface H asymptotic to the unperturbed horizon H0 with rate ρ. Theorem 3.4 identifies H with the event horizon of the perturbed spacetime in the subextremal Kerr–Newman case, conditional on an additional global far-field escape assumption (existence of R0); Remark 3.5 gives sufficient far-field decay conditions for that assumption. The abstract claims the event-horizon identification for Kerr(-Newman) without this qualification.","tokens_in":12965,"tokens_out":40705,"duration_ms":349699,"significance":"If the flagged issues are repaired, this is a valuable conceptual contribution: it recasts horizon regularity as a normal-sink/unstable-manifold statement and gives a short, transparent alternative proof of the recent Chen–Klainerman horizon-regularity theorem [CK24]. The reliance on [Hin21] and [HPS77] is a legitimate dependency, not a circularity, and the derivation is parameter-free. Theorem 3.2, the core result, is sound in substance: although the intended nullness reduction contains a circular step, the later decomposition argument (3.8) supplies a valid independent proof of nullness. The principal weaknesses are the unproved global escape condition behind Theorem 3.4, the abstract's unqualified statement of the event-horizon identification, and the deferred µ′(r+) < 0 (cosmological horizon) case.","major_comments":[{"comment":"The identity 'g(L, X̃) = g(L, X) = 0 at t∗ = t1' assumes that every X ∈ T_{z0}H on the slice t∗ = t1 is g-orthogonal to the null generator L; this is exactly the statement that T_{z0}H = L⊥, i.e. that H is null at that slice, the property the argument is meant to establish. For a hypersurface that is only known to be ruled by null geodesics, g(L, X) = 0 holds for X in a proper hyperplane of T_{z0}H, not for all X. The later argument 'H is null: conclusion' based on the decomposition (3.8) provides an independent and valid proof of nullness, so Theorem 3.2 survives, but the reduction step as written is circular and should be removed or replaced.","section":"Proof of Theorem 3.2, 'H is null: reduction'"},{"comment":"The conclusion that H is the event horizon is conditional on the assumed existence of R0 such that every point in {t∗ ≥ 0, r ≥ R0} can be connected to future null infinity by a future causal curve. This is a genuine additional global hypothesis: it is not derived from g − g0 ∈ ρC∞b (which controls decay as t∗ → ∞ at fixed r) nor from the local normal-sink dynamics, and the proof uses it exactly once to show E ⊂ J−(I+). Remark 3.5 gives sufficient far-field decay conditions for the R0 assumption, but it does not show that these conditions follow from the hypotheses of Theorem 3.4, and the theorem does not state them as hypotheses. The abstract's claim that in the Kerr(-Newman) case 'we show that H is equal to the boundary of the black hole region' is therefore stronger than the statement proved in Theorem 3.4; please either add an explicit far-field hypothesis to Theorem 3.4 and qualify the abstract accordingly, or prove the escape condition from natural asymptotic-flatness assumptions.","section":"Theorem 3.4 and Remark 3.5"},{"comment":"The abstract and the introduction list event and cosmological horizons of subextremal Kerr–Newman–de Sitter as examples to which the results apply, but the construction of §3 assumes µ′(r+) > 0 throughout, and the positivity of κ = µ′(r+)/(2b(r²+ + a²)) is used in the normal-sink estimate of Lemma 3.1 and in the sign structure of the sets B and E in the proof of Theorem 3.2. The footnote deferring the µ′(r+) < 0 (cosmological horizon) case to 'sign changes in the arguments below' does not establish the claimed scope, since the sign of µ′ flips the timelike/spacelike character of dr on the two sides of the horizon and exchanges the roles of the two sides. Please either work out the cosmological-horizon case or restrict the claims in the abstract and introduction.","section":"§3 footnote 1 and the abstract"}],"minor_comments":[{"comment":"In the case of a maximal future null geodesic γ with t∗ ◦ γ bounded and ¯r ≥ r+, the assertion that γ 'forces r(γ(s)) ≥ r+ + δ for some s' is not justified: the geodesic could in principle exit through r = r+ − 2δ while having previously reached values ≥ r+. A monotonicity or no-return argument (for example using that dr is timelike below r+ for large t∗) is needed to exclude this alternative.","section":"Proof of (3.8) in Theorem 3.2"},{"comment":"The proof works with B and E defined by (3.7) on the extended manifold X = [r+ − 2δ, ∞) × S² and invokes the decomposition (3.8), but (3.8) is proved in Theorem 3.2 only on the local spacetime with bounded r; the extension of the decomposition to the extended spacetime should be stated explicitly. Also, the assertion '(B ∪ H) ∩ J−(I+) = ∅' is made without proof; it follows from (3.8) together with E ⊂ J−(I+) and the fact that a causal curve reaching I+ witnesses membership in E, but the derivation should be spelled out, as should the derivation of items (1)–(3) of the theorem.","section":"Proof of Theorem 3.4"},{"comment":"The symbol X is used both for the physical space S² × (r+ − 2δ, r+ + 2δ) in the statement of Theorem 3.2 and for the phase space R_ˆσ × (T∗S²) in (3.5) and in the definition of the map F in the proof; this makes the proof of Theorem 3.2 very hard to follow, and distinct notation should be used for the two spaces.","section":"§3, especially (3.5) and the proof of Theorem 3.2"},{"comment":"In the statement of Theorem 3.4, B and E are defined as the components of (M ∩ {t∗ ≥ t0}) \\ H on which r is bounded, resp. unbounded, while in the proof they are redefined by (3.7) with t0 in place of t1; the identification of the two definitions is part of the conclusion. Please clarify this by using separate symbols or by stating explicitly that the (3.7)-defined sets are provisional and their equality with the components is established in the proof.","section":"Statement of Theorem 3.4"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a short note whose core theorem (Theorem 3.2) is a clean application of the author's earlier normally hyperbolic framework [Hin21]; the dependency is legitimate and clearly credited, and the novelty lies in the reformulation and in the application to horizon regularity. My main concerns, also reflected in the major comments, are that (i) the intended nullness proof in Theorem 3.2 is circular as written, though a valid independent argument appears later in the same proof, and (ii) the abstract and introduction promise more than the theorems deliver, since the event-horizon identification is conditional on a far-field escape condition and the cosmological-horizon case is deferred. Both issues are repairable in a revision. The paper is well suited to a mathematical relativity journal, and its brevity is appropriate for a note."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing you should know: this is a worthwhile paper, and the local horizon-construction part holds up. Hintz recasts Chen–Klainerman's Kerr horizon regularity as an application of a general unstable manifold theorem for maps that translate in time and have a normal sink at an invariant manifold. The machinery is genuinely general—Theorem 2.2 is stated abstractly and proved by reduction to [Hin21, Theorem 2.3] plus an invariant-subbundle argument from [HPS77]. Then he checks the normal-sink condition for the projected null geodesic flow near subextremal Kerr–Newman horizons (Lemma 3.1) and gets existence and uniqueness of a smooth null hypersurface asymptotic to the stationary horizon (Theorem 3.2). That is a real step beyond Chen–Klainerman, which treats subextremal Kerr only, and the proof is honest: it cites the heavy PDE machinery rather than re-deriving it.\n\nWhere the paper is softest is Theorem 3.4, the event-horizon identification. The local theorem proves existence of H, but showing H equals the boundary of J^-(I+) needs a global far-field escape condition: every point far outside (r >= R0) must be able to reach future null infinity. That condition is assumed, not derived from the local decay hypotheses. Remark 3.5 gives sufficient conditions, but they are exactly that—extra assumptions. If they fail, H may still be a smooth null hypersurface in the local sense while not being the event horizon. So the advertised physical conclusion is conditional. This is a presentation gap rather than a discovered error, but it should be fixed in revision: either state Theorem 3.4 with the escape condition as an explicit hypothesis (it nearly does) or prove it from stronger decay/global geometric hypotheses.\n\nMinor: the de Sitter/cosmological-horizon case, where mu'(r+) < 0, is only sketched as 'track the sign changes.' That is fine for a note, but it should be written out or explicitly deferred. Also, there is a small unjustified step in the nullness reduction where an arbitrary X in T H at t* = t1 is asserted to have g(L,X) = 0; the later B/E decomposition seems to give an independent route to nullness, so this does not threaten Theorem 3.2.\n\nBottom line: the central local theorem is solid and novel; the event-horizon statement is honest but conditional. This deserves peer review. A serious referee should focus on Theorem 3.4's hypotheses and the sign-case write-up. I would bring it to a reading group and would cite the abstract unstable manifold theorem.","headline":"A clean, useful extension of Chen–Klainerman's horizon-regularity result via a general unstable manifold theorem; the local result is solid, and the event-horizon identification is honest but conditional on a global escape assumption that the paper does not derive.","tokens_in":13529,"tokens_out":2328,"would_cite":true,"duration_ms":22733,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37C70","83C57","53B50"],"pacs":[],"model":"deepseek-v4-flash","headline":"For spacetimes settling to a stationary black hole, a unique smooth horizon emerges at late times.","keywords":["event horizon","null hypersurface","unstable manifold","normal sink","Kerr–Newman spacetime","asymptotically stationary spacetimes","null-geodesic flow","dynamical black holes"],"falsifier":"Construct an asymptotically stationary metric $g=g_0+h$ with $h\\in\\rho C^\\infty_b$ and a compact far-field shell that traps null geodesics, so that some points at arbitrarily large radius cannot reach future null infinity while the metric outside the shell is exactly Kerr–Newman. If Theorem 3.4's conclusion is correct, the unique $\\mathcal{H}$ from Theorem 3.2 would still have to bound the black hole region, whereas the engineered escape failure would make the true event horizon deviate from $\\mathcal{H}$.","tokens_in":12461,"feed_emoji":"🕳️","tokens_out":5648,"duration_ms":51528,"temperature":0.7,"pith_summary":"This paper proves that a dynamical spacetime which decays to a stationary spacetime containing a horizon $\\mathcal{H}_0$ must contain a unique smooth null hypersurface $\\mathcal{H}$ asymptotic to $\\mathcal{H}_0$. The result is cast as a general unstable-manifold theorem for perturbations of flows that translate in time and have a normal sink at an invariant manifold. For asymptotically subextremal Kerr–Newman spacetimes, the paper further shows that $\\mathcal{H}$ is exactly the boundary of the black hole region, i.e. the future event horizon. A sympathetic reader cares because it turns a global, hard-to-access object—the event horizon—into a late-time local object that is uniquely determined by the asymptotic metric alone.","feed_headline":"Unique smooth horizon emerges for perturbed black holes","feed_subtitle":"Spacetimes settling to Kerr–Newman develop a unique smooth horizon that is the true event horizon.","key_machinery":"The machinery is a general unstable-manifold theorem for maps that shift time by one and contract the normal bundle at an invariant submanifold. In the spacetime application, the invariant manifold is the conormal bundle of the unperturbed horizon, viewed in the spherical cotangent bundle, and the contracting map is the time-one null-geodesic flow. The key mechanism is that the linearization of the null-geodesic vector field along the horizon has a strictly negative normal Lyapunov exponent, making the horizon a normal sink; the unstable-manifold theorem then produces a unique invariant manifold carrying the perturbed horizon, and a geometric argument shows its base projection is null and ruled by null geodesics.","core_discovery":"The central discovery is Theorem 3.2: given a Lorentzian metric $g$ differing from a stationary Kerr–Newman-type metric $g_0$ by a term in $\\rho C^\\infty_b$, there is a $t_0$ and a unique smooth null hypersurface $\\mathcal{H}\\subset\\{t_*\\ge t_0\\}$ that approaches the unperturbed horizon $\\mathcal{H}_0$ in a $\\rho C^\\infty_b$ sense and is ruled by null geodesics. Theorem 3.4 adds that, under a mild global far-field condition, $\\mathcal{H}$ coincides with the event horizon of $(M,g)$, defined as the boundary of the causal past of future null infinity. The argument constructs $\\mathcal{H}$ as the base projection of an unstable manifold in the spherical cotangent bundle, obtained from the phase space of the null-geodesic flow; the horizon is therefore produced by a contraction mechanism, not by a global causal analysis.","pith_inferences":["Beyond the paper: the same unstable-manifold mechanism may identify other distinguished null hypersurfaces—for instance inner horizons or Cauchy horizons—whenever the normal contraction is replaced by the appropriate expansion.","Beyond the paper: the uniqueness statement suggests that horizon extraction at late times is structurally stable; any two numerical or analytical procedures that track the apparent horizon should converge to the same hypersurface once the spacetime settles.","A testable extension would replace the $C^\\infty_b$ decay weight by weaker or non-smooth weights and ask whether the perturbed horizon inherits exactly the regularity of the perturbation rather than remaining smooth.","The far-field escape condition of Theorem 3.4 could be probed numerically by constructing a metric with a localized shell that traps null geodesics, then checking whether the locally unique $\\mathcal{H}$ still equals the boundary of $J^-(\\mathcal{I}^+)$."],"forward_implications":["If $g$ decays to any stationary Kerr–Newman-type metric with a non-degenerate horizon, the late-time horizon $\\mathcal{H}$ exists and is unique, independent of gauge or foliation choices.","In subextremal Kerr and Kerr–Newman, $\\mathcal{H}$ is the actual event horizon, so the event horizon of a perturbed black hole is smooth rather than merely Lipschitz.","The same construction covers event and cosmological horizons of subextremal Kerr–Newman–de Sitter spacetimes, with the analogous conclusion relative to the conformal boundary.","If the perturbation is small in $\\rho C^2_b$, the horizon can be constructed from $t_0=0$, i.e. on the entire future.","Under the global far-field assumption, every future causal curve starting on the far side of $\\mathcal{H}$ escapes to null infinity, while curves starting on the other side stay in the black hole region."],"supporting_citations":[{"why":"proves regularity of the future event horizon in perturbations of Kerr, the result this paper generalizes and reproves by a shorter argument.","marker":"[CK24]"},{"why":"supplies the normally hyperbolic trapping and unstable-manifold framework on asymptotically stationary spacetimes that the proof adapts.","marker":"[Hin21]"},{"why":"provides the invariant-manifold theorem (their Theorem 3.5) used to produce the invariant normal-bundle section.","marker":"[HPS77]"},{"why":"contains the null-geodesic flow analysis near horizons whose normal-sink argument is refined in Lemma 3.1.","marker":"[Vas13]"},{"why":"gives the generalized normal contraction estimate near horizons used for the sink condition.","marker":"[Gan19]"},{"why":"supplies the explicit Kerr–Newman(–de Sitter) metric form in Boyer–Lindquist coordinates used in Section 3.","marker":"[PG06]"},{"why":"is one of the nonlinear stability works whose spacetimes satisfy Theorem 3.4's far-field condition, showing the global assumption is realistic.","marker":"[DHRT21]"},{"why":"is another nonlinear stability result, for Kerr with small angular momentum, supplying spacetimes in which $\\mathcal{H}$ equals the event horizon.","marker":"[KS23]"}],"fun_headline_variants":["Unique horizon guaranteed in settling spacetimes","Perturbed Kerr-Newman: unique horizon exists","Smooth horizon emerges after black hole settles","Existence of unique horizon for dynamical spacetimes","Horizon proven for asymptotic stationary spacetimes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the global condition in Theorem 3.4 that every sufficiently distant point can be joined to future null infinity by a future causal curve; if that fails, the locally constructed $\\mathcal{H}$ need not be the boundary of $J^-(\\mathcal{I}^+)$, only a unique asymptotic null hypersurface.","fun_headline_variants_meta":{"raw":{"variants":["Unique horizon guaranteed in settling spacetimes","Perturbed Kerr-Newman: unique horizon exists","Smooth horizon emerges after black hole settles","Existence of unique horizon for dynamical spacetimes","Horizon proven for asymptotic stationary spacetimes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000243,"raw_usage":{"total_tokens":1492,"prompt_tokens":872,"completion_tokens":620,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":488,"completion_tokens_details":{"reasoning_tokens":548}},"tokens_in":488,"tokens_out":620,"duration_ms":6616,"temperature":1.0,"reasoning_tokens":548,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T17:23:00.921210+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Construct an asymptotically stationary metric $g=g_0+h$ with $h\\in\\rho C^\\infty_b$ and a compact far-field shell that traps null geodesics, so that some points at arbitrarily large radius cannot reach future null infinity while the metric outside the shell is exactly Kerr–Newman. If Theorem 3.4's conclusion is correct, the unique $\\mathcal{H}$ from Theorem 3.2 would still have to bound the black hole region, whereas the engineered escape failure would make the true event horizon deviate from $\\mathcal{H}$.","supporting_citations":[],"review_version":1}