Pipeline-generated

How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

Trapped surfaces and Penrose incompleteness

1 · Prerequisites

2 · Summary

3 · Logical flowchart

4 · Definitions, postulates, theorems and proofs

DefinitionDefinition: Literature-sourcedProof: Not applicableverified 2026-10-22 (gpt-6.1-sol)Open item page →

Future null geodesic completeness and incompleteness

Physical scope

A purely geometric definition of affine completeness within a specified Lorentzian manifold; it asserts neither curvature blow-up nor the existence of a singular point.

Definition

Let $(M, g, N)$ be a time-oriented smooth Lorentzian four-manifold with metric-selected connection $\nabla$ . A future-directed affinely parametrized null geodesic is a smooth nonconstant curve $γ : J \to M$ , where $J \subseteq R$ is an interval with nonempty interior, such that its tangent $\dot{γ}$ is future-directed and null at every interior parameter and $D_{λ} \dot{γ} = 0$ . Here “null” means $g (\dot{γ}, \dot{γ}) = 0$ and $\dot{γ} \neq 0$ , and future-directed means $g (\dot{γ}, N) < 0$ . The geodesic equation and covariant derivative are those of the metric-selected connection.

The maximal future affine interval of such a geodesic is the future endpoint interval of its maximal extension as an affinely parametrized geodesic in $M$ . It is future complete when that interval has no finite upper endpoint, equivalently when its affine parameter tends to $+ \infty$ . The spacetime is future null geodesically complete when every future-directed null geodesic is future complete, and future null geodesically incomplete when at least one is not.

An affine change of parameter $\tilde{λ} = a λ + b$ with $a > 0$ preserves whether the maximal future interval is unbounded above, so these properties do not depend on the chosen positive affine normalization.

Definition references

Geodesic of an affine connection

Causal curves, chronological and causal reachability, and inextendibility

Metric compatibility, torsion freedom and the metric-selected connection class

Smooth time-oriented Lorentzian four-manifolds: the geometric objects of the model

Conventions and scope

This definition distinguishes an incomplete geodesic in $M$ from a curvature singularity or a point added to an extension. No equation for matter, Einstein equation, or claim about the physical adequacy of $(M, g, N)$ is included.

LemmaStatement: AI-adaptedProof: AI-generatedprecheck passverified 2026-10-22 (gpt-6.1-sol)Open item page →

A complete temporal flow identifies smooth Cauchy slices

Physical scope

A differential-topological statement about two smooth Cauchy hypersurfaces and a supplied complete timelike product flow; no field equation or empirical claim is involved.

Statement

Let $(M, g, N)$ be a time-oriented smooth Lorentzian four-manifold. Suppose a smooth three-manifold $S_{0}$ and a smooth diffeomorphism $Ψ : R \times S_{0} \to M$ are supplied such that each curve $s \mapsto Ψ (s, z)$ is an inextendible future timelike curve and $Ψ (0, z) = z$ . Let $Σ \subset M$ be a smooth spacelike Cauchy hypersurface, meaning that it is met exactly once by every inextendible causal curve. Then the second-coordinate projection $P : = {pr}_{S_{0}} \circ Ψ^{- 1} ∣_{Σ} : Σ \to S_{0}$ is a diffeomorphism. In particular, $Σ$ and $S_{0}$ have the same connectedness and compactness properties.

Statement references

Spacelike hypersurfaces, acausal sets, Cauchy sets, domains of dependence and Cauchy horizons

Causal curves, chronological and causal reachability, and inextendibility

Diffeomorphisms and local diffeomorphisms of manifolds

Smooth time-oriented Lorentzian four-manifolds: the geometric objects of the model

The smooth inverse function theorem on manifolds

Facts & Assumptions

Given: A complete timelike product flow $Ψ$ as in the Statement and a smooth spacelike Cauchy hypersurface $Σ$ .

[F1]

The map $Ψ$ is a diffeomorphism, each $s$ -curve is an inextendible future timelike curve, and $Ψ (0, z) = z$ ; these are the explicit product-flow hypotheses. In Causal curves, chronological and causal reachability, and inextendibility, timelikeness of a curve requires a future timelike velocity almost everywhere.

[F2]

Spacelike hypersurfaces, acausal sets, Cauchy sets, domains of dependence and Cauchy horizons: a Cauchy hypersurface meets every inextendible causal curve exactly once, and a smooth spacelike hypersurface has a three-dimensional positive-definite tangent space at each point.

[F3]

The smooth inverse function theorem on manifolds: a smooth map with an isomorphism as differential at a point restricts to a diffeomorphism on suitable open neighborhoods.

[F4]

Write $V (s, z) = d Ψ_{(s, z)} (\partial_{s})$ . It is nonzero because $d Ψ$ is invertible. For each fixed $z$ , the smooth function $s \mapsto g (V (s, z), V (s, z))$ is negative almost everywhere by [F1], hence is nonpositive everywhere: a positive value would, by continuity, give a positive value on an interval of positive measure. Thus every nonzero vector in the orbit tangent line has nonpositive squared norm, whereas [F2] gives positive squared norm for every nonzero vector tangent to $Σ$ . Their intersection is therefore $0$ , even at an exceptional null orbit tangent.

Proof

Given: [F1]–[F4].

1.1F1F2givenconstruct

For each $z \in S_{0}$ , the orbit $s \mapsto Ψ (s, z)$ is an inextendible timelike curve, so [F2] says it meets $Σ$ in exactly one point. Define $P (p)$ by writing $Ψ^{- 1} (p) = (s, z)$ and setting $P (p) = z$ . Every $z$ occurs because its orbit meets $Σ$ , and if $P (p) = P (q)$ then $p, q$ lie on the same orbit, which meets $Σ$ only once; hence $P$ is bijective.

1.2F1given

The map $P$ is smooth because it is the restriction of the smooth map ${pr}_{S_{0}} \circ Ψ^{- 1}$ to the embedded submanifold $Σ$ .

1.3F1F2F4givenalgebra

At $p = Ψ (s, z) \in Σ$ , the kernel of $d ({pr}_{S_{0}} \circ Ψ^{- 1})_{p}$ is the one-dimensional tangent to the $s$ -orbit. If $v \in T_{p} Σ$ is in the kernel, then $v$ lies in that orbit tangent line and is tangent to the spacelike hypersurface. By [F4], $v = 0$ . Since both tangent spaces have dimension three, $d P_{p} : T_{p} Σ \to T_{z} S_{0}$ is an isomorphism.

2.1F3step 1.1step 1.2step 1.3∎

By [F3], $P$ is a local diffeomorphism at every point. It is already a bijection by step 1.1, so its local smooth inverses agree with the unique set-theoretic inverse $P^{- 1}$ on their domains; these neighborhoods cover $S_{0}$ , proving that $P^{- 1}$ is smooth. Thus $P$ is a diffeomorphism.

Remarks

The proof uses the Cauchy property of $Σ$ to identify each flow orbit once, and spacelikeness to make the projection a local diffeomorphism. It does not assert that an arbitrary continuous Cauchy set is smoothly diffeomorphic to a chosen slice.

DefinitionDefinition: Literature-sourcedProof: Not applicableverified 2026-10-22 (gpt-6.1-sol)Open item page →

Trapping and the asymptotic black-hole region

Physical scope

Pure Lorentzian-geometric definitions. No field equation, cosmic-censorship assumption, event-horizon existence, or empirical claim is asserted.

Definition

Let $(M, g, N)$ be a smooth time-oriented Lorentzian four-manifold of signature $(-, +, +, +)$ . Let $S \subset M$ be a nonempty smooth embedded two-manifold, compact without boundary, whose induced metric $q$ is positive definite. At each point its normal plane $(T_{p} S)^{⊥}$ is a Lorentzian two-plane with two future null rays. On any open patch carrying a smooth ordered pair of future null normals $ℓ, k$ with $g (ℓ, k) = - 2$ , define $χ_{ℓ} (X, Y) = g (\nabla_{X} ℓ, Y), θ_{ℓ} = {tr}_{q} χ_{ℓ}, χ_{k} (X, Y) = g (\nabla_{X} k, Y), θ_{k} = {tr}_{q} χ_{k}$ for tangent vectors $X, Y$ to $S$ . These are the null second fundamental forms and their expansions. A future trapped surface is such an $S$ for which, on every local patch and for both future null normal rays, the corresponding expansion is strictly negative at every point. Positive rescaling of a null normal multiplies its expansion by the same positive factor, so the sign condition is independent of its affine normalization and of a local interchange of the two rays.

When a specified conformal completion is supplied, write it as $(\overline{M}, \overline{g}, \overline{N}, Ω, j, I^{+})$ . Here $\overline{M}$ is a smooth four-dimensional manifold with boundary in the senses of Topological manifolds with boundary and Smooth charts, atlases, and structures with boundary. At each point $\overline{g}$ is a symmetric bilinear form on the tangent space, its components extend smoothly in every boundary chart, and its pointwise inertia is $(3, 1, 0)$ in the convention of Lorentz forms of inertia (3,1,0) on a 4-dimensional real vector space. The vector field $\overline{N}$ likewise has components that extend smoothly in every boundary chart; it satisfies $\overline{g} (\overline{N}, \overline{N}) < 0$ everywhere and $\overline{g} (\overline{N}, j_{*} N) < 0$ on the interior. The map $j : M \to {\overline{M}}^{\circ}$ is a diffeomorphism onto the interior. The function $Ω : \overline{M} \to R$ is smooth in the boundary-chart sense, positive on $j (M)$ , and zero with nonzero differential on the designated smooth three-dimensional boundary subset $I^{+} \subseteq \partial \overline{M}$ . On $j (M)$ the metric satisfies $\overline{g} ∣_{j (M)} = Ω^{2} (j^{- 1})^{*} g$ , where pullback is as in The pullback of a covariant tensor field. The restriction of $\overline{g}$ to $T I^{+}$ has a one-dimensional kernel at every point; this null line is the generator direction, and its future ray is the one whose vectors $v$ satisfy $\overline{g} (v, \overline{N}) < 0$ . Thus the designated boundary is null and future-oriented; no completeness of its generators is assumed.

A curve $γ : [0, 1] \to \overline{M}$ is locally absolutely continuous when it is continuous and, around each parameter value, its coordinate functions in a boundary or interior chart are absolutely continuous on every compact subinterval of that chart neighborhood, in the sense of Absolute continuity on a compact interval. Boundary charts take values in relatively open subsets of the closed half-space. To compare two charts around a parameter value, shrink its parameter neighborhood so the coordinate image lies in a compact convex subset of a local-extension domain of the transition map. Its derivative is bounded there, so the mean-value estimate (The mean value theorem, as the case $g (x) = x$ of Cauchy's: for $f$ continuous on $[a, b]$ with $a < b$ and differentiable on $(a, b)$ there is $c \in (a, b)$ with $f (b) - f (a) = f^{'} (c) (b - a)$ ) makes the transition Lipschitz. Applying that bound to the finite disjoint-interval condition for absolute continuity shows that the transformed coordinate functions remain absolutely continuous; the argument in both directions proves chart independence. At an interior parameter value where the coordinate derivatives exist, $\dot{γ} (t)$ is the tangent vector with those components; the chain rule for the chart transitions (Chain rule for smooth half-space maps) makes it independent of chart. A future causal curve in $\overline{M}$ is a locally absolutely continuous curve for which, for almost every $t \in (0, 1)$ , this derivative exists and satisfies $\overline{g} (\dot{γ}, \dot{γ}) \leq 0$ and $\overline{g} (\dot{γ}, \overline{N}) < 0$ . These inequalities make the derivative nonzero and select the future direction. On the interior, $\overline{g} (j_{*} v, j_{*} v) = Ω^{2} g (v, v)$ for every tangent vector $v$ , so causal character is unchanged, and the specified orientation condition makes the future cone agree with that of $N$ (Smooth time-oriented Lorentzian four-manifolds: the geometric objects of the model). Define ${\overline{J}}_{\overline{g}, \overline{N}}^{-} (I^{+}) : = {x \in \overline{M} : there are y \in I^{+} and a future causal curve in \overline{M} from x to y}$ . The black region relative to this completion and its event horizon are, respectively, $B_{I^{+}} : = {p \in M : j (p) \notin {\overline{J}}_{\overline{g}, \overline{N}}^{-} (I^{+})}, H_{I^{+}} : = \partial_{M} B_{I^{+}} .$ If no such completion is supplied, these two sets are undefined; their existence is not a consequence of the definition of a trapped surface.

Given in addition a designated smooth future null normal ray field $[ℓ_{o u t}]$ on $S$ , called the outward direction, $S$ is marginally outer trapped relative to this choice when $θ_{ℓ_{o u t}} = 0$ everywhere. If the outward label comes from a side of an embedded spacelike hypersurface or from an asymptotic end, that side or end is part of the supplied data; it is not inferred from $S$ alone. The zero condition is invariant under positive rescaling of the chosen ray.

Definition references

Null normal pairs and screen geometry

Causal curves, chronological and causal reachability, and inextendibility

Smooth time-oriented Lorentzian four-manifolds: the geometric objects of the model

Lorentz forms of inertia (3,1,0) on a 4-dimensional real vector space

Absolute continuity on a compact interval

$C^{r}$ and smooth maps between smooth manifolds

The differential of a smooth map

The pullback of a covariant tensor field

Derivations at a point and the tangent space

Diffeomorphisms and local diffeomorphisms of manifolds

Intervals of $R$ : the nine order-convex forms, nondegeneracy, and length

Chain rule for smooth half-space maps

The mean value theorem, as the case $g (x) = x$ of Cauchy's: for $f$ continuous on $[a, b]$ with $a < b$ and differentiable on $(a, b)$ there is $c \in (a, b)$ with $f (b) - f (a) = f^{'} (c) (b - a)$

Smooth functions on relatively open half-space sets

Topological manifolds with boundary

Smooth charts, atlases, and structures with boundary

Interior, closure, boundary, exterior, derived set and isolated point in a topological space

Conventions and scope

The screen geometry supplier fixes the null second fundamental form and expansion and proves their boost law: under $ℓ \mapsto e^{f} ℓ$ , one has $θ_{ℓ} \mapsto e^{f} θ_{ℓ}$ . Thus the two inequalities in the trapped-surface definition are local, branch-independent conditions. The event horizon is a global boundary in a specified completion; it is not identified with a trapped surface or a marginally outer trapped surface. In the SI geometric convention $x^{0} = c t$ with dimensionless metric and affine parameter in metres, the expansion has units $m^{- 1}$ , as in the screen geometry supplier.

LemmaStatement: Literature-sourcedProof: AI-generatedprecheck passverified 2026-10-22 (gpt-6.1-sol)Open item page →

Compactness of the future horismos under null completeness

Physical scope

A conditional Lorentzian-geometric deduction from compact trapping, global hyperbolicity, null Ricci convergence, and future null completeness; it asserts no curvature singularity or event horizon.

Statement

Assume the Axioms of Countable Choice and Dependent Choice. Let $(M, g, N)$ be a time-oriented smooth Lorentzian four-manifold that is globally hyperbolic, meaning strongly causal with compact causal diamonds. Let $S \subset M$ be a nonempty compact smooth embedded spacelike two-surface whose two future null expansions are negative everywhere. Assume the direct null Ricci condition $Ric (ℓ, ℓ) \geq 0$ for every null vector $ℓ$ and assume every future-directed affinely parametrized null geodesic in $M$ is future complete. Define $I^{+} (S) = ⋃_{s \in S} I^{+} (s), J^{+} (S) = ⋃_{s \in S} J^{+} (s), E^{+} (S) = J^{+} (S) ∖ I^{+} (S) .$ Then $E^{+} (S) = \partial I^{+} (S)$ is a closed topological three-manifold without boundary and is compact in the subspace topology of $M$ .

Statement references

Trapping and the asymptotic black-hole region

Chronology, causality, strong causality, global hyperbolicity and time functions

Causal curves, chronological and causal reachability, and inextendibility

Future null geodesic completeness and incompleteness

Null Raychaudhuri and focal-point bound

Null focal points exclude later horismos generators

Facts & Assumptions

Given: A time-oriented globally hyperbolic smooth Lorentzian four-manifold, a nonempty compact future trapped surface $S$ , null Ricci nonnegativity, Countable Choice, Dependent Choice, and future null geodesic completeness.

[F1]

Chronology, causality, strong causality, global hyperbolicity and time functions: global hyperbolicity is strong causality together with compactness of each causal diamond $J^{+} (p) \cap J^{-} (q)$ .

[F2]

Causal curves, chronological and causal reachability, and inextendibility: $p ≪ q$ means a future timelike curve joins $p$ to $q$ ; $p \leq q$ means $p = q$ or a future causal curve joins them; $I^{+} (S)$ and $J^{+} (S)$ are the corresponding unions over $s \in S$ .

[F3]

In a globally hyperbolic manifold the causal relation is closed and truncated causal cones over compact sets are compact: in a globally hyperbolic spacetime, $J^{+} (K)$ is closed for every compact $K \subseteq M$ .

[F4]

Time separation in a globally hyperbolic manifold: finiteness, existence of maximizing geodesics and the radial formula on causally convex normal regions: if $s \leq p$ and $s \neq p$ , a maximizing causal curve from $s$ to $p$ exists and is a causal geodesic up to increasing reparametrization; its Lorentzian length is zero exactly when $s ≪̸ p$ .

[F5]

Covariant differentiation along curves: metric pairing, parallel transport and geodesic speed: along an affinely parametrized geodesic, $g (\dot{γ}, \dot{γ})$ is constant.

[F6]

Future sets, past sets, achronal boundaries and their elementary properties and Openness of the chronological future and past, openness of the chronological relation, and closure of future and past sets: $I^{+} (S)$ is an open future set, so its boundary is an achronal boundary.

[F7]

Achronal boundaries are locally Lipschitz graphs, topological hypersurfaces and null for smooth volume densities: the boundary of an open future set is closed and is a topological three-manifold without boundary. Its statement assumes Countable Choice and Dependent Choice.

[F8]

Null normal pairs and screen geometry: on a patch with a smooth future null normal $ℓ$ , $θ_{ℓ} = {tr}_{q} χ_{ℓ}$ , where $χ_{ℓ} (X, Y) = g (\nabla_{X} ℓ, Y)$ ; a positive boost $ℓ \mapsto e^{f} ℓ$ sends $θ_{ℓ}$ to $e^{f} θ_{ℓ}$ .

[F9]

Local null normal pairs and the positive screen quotient: near every point of a spacelike two-surface, there are smooth future null normal fields representing the two rays, normalized as a pair by $g (ℓ, k) = - 2$ .

[F10]

Null Raychaudhuri and focal-point bound: along a smooth null normal congruence with $Ric (ℓ, ℓ) \geq 0$ and $θ (0) < 0$ , a focal instant occurs by affine parameter $2 / ∣ θ (0) ∣$ if the congruence exists through that parameter.

[F11]

Null focal points exclude later horismos generators: a focal instant on a future null geodesic normal to $S$ puts all sufficiently close later points on that generator in $I^{+} (S)$ , provided the geodesic and variation data continue slightly past the focal instant.

[F12]

The geodesic initial-value problem and the exponential map of the metric-selected connection: the maximal geodesic evaluation $(t, p, v) \mapsto γ_{p, v} (t)$ is smooth on its open domain ${(t, p, v) : t \in I_{p, v}}$ , where $I_{p, v}$ is the maximal affine interval for initial data $(p, v)$ . The exponential map satisfies $\exp_{p} (t v) = γ_{p, v} (t)$ when defined, its differential at zero is the identity, and short geodesics are its radial curves.

[F13]

The joint exponential map: smoothness, invertible differential at the diagonal, and the small-velocity two-point inverse: the joint exponential map is a diffeomorphism near $(s, 0)$ with differential $(X, V) \mapsto (X, X + V)$ ; its inverse therefore has differential $(X, Y) \mapsto (X, Y - X)$ on the diagonal.

[F14]

Existence of strictly convex normal neighborhoods: small coordinate balls are normal neighborhoods of each of their points and Local causality in a convex normal ball: causal reachability is the exponential cone, the Lorentzian length is bounded by the displacement radius, and null endpoints force the null geodesic: in a sufficiently small strictly convex normal neighborhood, a future timelike position vector gives a future timelike radial geodesic.

[F15]

The push-up rule: p << q and q <= r imply p << r, together with the timelike-piece criterion: if $s \leq r$ and $r ≪ p$ , then $s ≪ p$ ; a future causal curve containing a timelike subcurve has chronologically related endpoints.

[F16]

Smooth dependence of ODE solutions on parameters: for a smooth ODE, near fixed initial data there is a common local compact time interval on which solutions depend smoothly on the initial data and parameters. For the full maximal geodesic domain use [F12].

[F17]

A continuous image of a compact space is compact; a continuous real-valued map on a nonempty compact space attains a maximum and a minimum; and a continuous bijection from a compact space to a Hausdorff space is a homeomorphism, claim 1: for topological spaces $X, Y$ and a continuous map $f : X \to Y$ , the image $f [K]$ of every compact subset $K \subseteq X$ is compact in $Y$ .

[F18]

A closed subspace of a compact space is compact, and a finite union of compact subspaces is compact: a closed subset of a compact topological space is compact, and a finite union of compact subsets of any topological space is compact.

[F19]

Heine-Borel in $R^{n}$ : with the Euclidean metric a subset of $R^{n}$ is compact if and only if it is closed and bounded, and the proof by bisection uses no choice principle; the same holds on the real line and For a metric space with its metric topology, compactness in the topological sense is compactness in the metric sense, and the two notions of compact subset coincide: closed bounded Euclidean subsets, including closed balls and $[0, L]$ , are compact for their metric topologies; compactness in a metric space agrees with topological compactness for that topology.

[F20]

Smooth time-oriented Lorentzian four-manifolds: the geometric objects of the model: the time-orientation field $N$ is future timelike, and a future null vector $ℓ$ satisfies $g (ℓ, N) < 0$ .

[F21]

Future null geodesic completeness and incompleteness: future null completeness means every maximal future affine interval of a future null geodesic is unbounded above.

[F22]

A product of finitely many compact spaces is compact in the product topology: the product of finitely many compact spaces is compact.

[A1]

Countable Choice and Dependent Choice are explicit mathematical assumptions, as defined by The Axiom of Countable Choice ( ${A C}_{ω}$ ) and The axiom of dependent choice: a relation in which every element is related to something admits an $N$ -indexed chain.

[A2]

$S$ is future trapped and compact; the null Ricci inequality holds along every future null normal geodesic; and every such geodesic exists for all future affine parameters. These are hypotheses, not deductions from an Einstein equation or a matter-energy condition.

Proof

Given: [F1]–[F22] and [A1]–[A2].

1.1F2F6givenconstruct

Put $F : = I^{+} (S)$ . It is open because it is the union of the open sets $I^{+} (s)$ , $s \in S$ . If $x \in F$ and $x ≪ y$ , then some $s \in S$ satisfies $s ≪ x$ ; transitivity gives $s ≪ y$ , so $y \in F$ . Thus $F$ is an open future set.

1.2A2F8F9F20givenchoose

For each point of $S$ , choose a neighborhood with the two smooth null normal branches supplied by [F9]. On each branch rescale by a positive smooth factor so that $g (ℓ, N) = - 1$ ; this is possible by [F20] and does not change the sign of the expansion by [F8]. Both normalized expansions are continuous and negative at the chosen point by the trapped-surface hypothesis. Shrink the neighborhood so that both are at most $- c_{j}$ for some $c_{j} > 0$ . Compactness of $S$ gives a finite subcover; the minimum of its finitely many positive constants gives $c > 0$ such that every normalized future null normal has initial expansion $θ_{0} \leq - c$ .

2.1F1F2F3F12F14F15F20step 1.1given

By [F1] and [F3], $J^{+} (S)$ is closed, and $F \subseteq J^{+} (S)$ ; hence $\overline{F} \subseteq J^{+} (S)$ . Therefore $\partial F = \overline{F} ∖ F \subseteq E^{+} (S)$ . Conversely, take $p \in E^{+} (S)$ and choose $s \in S$ with $s \leq p$ . The timelike vector $N_{p}$ and the local exponential map give future timelike points $p_{ϵ} \to p$ by [F12], [F14] and [F20]. If $s = p$ then $s ≪ p_{ϵ}$ ; if $s \neq p$ then $s \leq p ≪ p_{ϵ}$ and [F15] gives $s ≪ p_{ϵ}$ . Thus $p_{ϵ} \in F$ and, since $p \notin F$ , $p \in \partial F$ . Hence $E^{+} (S) = \partial F$ .

2.2A2F10F11F12F16F21step 1.2given

Set $L : = 2 / c$ . For each normalized null normal at $s \in S$ , future completeness [F21], local smooth dependence [F16], and smooth evaluation on the maximal geodesic domain [F12] give a smooth null normal geodesic congruence near that initial datum through $[0, L]$ . Applying [F10] with $θ_{0} \leq - c$ and the null Ricci hypothesis gives a first focal parameter $λ_{f} \leq 2 / ∣ θ_{0} ∣ \leq L$ . The extension required in [F11] is supplied by future null completeness, so its local variation data also exist slightly beyond $λ_{f}$ .

3.1A1F7step 2.1

Apply [F7] to the open future set $F$ . The equality in step 2.1 shows that $E^{+} (S)$ is closed and is a topological three-manifold without boundary.

3.2F2F4F5F20step 2.1given

Let $p \in E^{+} (S)$ and choose $s \in S$ with $s \leq p$ . If $s = p$ , use the zero-length segment at $s$ . If $s \neq p$ , [F4] supplies a maximizing causal geodesic $γ$ from $s$ to $p$ . Since $p \notin I^{+} (S)$ , $s ≪̸ p$ , so the maximizing length is zero; the nonconstant affine geodesic has constant squared speed by [F5], and therefore its future causal tangent is null. Rescale its affine parameter by a positive constant so that $g (\dot{γ} (0), N_{s}) = - 1$ , and translate the parameter so that its initial value is $0$ .

4.1F2F15step 2.1step 3.2

Every point $r$ on this segment satisfies $s \leq r \leq p$ . If $r \in I^{+} (S)$ , some $s^{'} \in S$ satisfies $s^{'} ≪ r \leq p$ ; the push-up rule [F15] then gives $s^{'} ≪ p$ , contradicting $p \notin I^{+} (S)$ . Thus the entire segment lies in $E^{+} (S)$ .

5.1F13F14F15step 3.2step 4.1givencontradiction

The initial null tangent $ℓ = \dot{γ} (0)$ is normal to $S$ . Suppose otherwise. Then some $X \in T_{s} S$ has $g (ℓ, X) \neq 0$ ; since $S$ is spacelike, $g (X, X) > 0$ . Choose a smooth curve $η (u) \subset S$ with $η (0) = s$ and $η^{'} (0) = X$ . The joint exponential map lemma [F13] gives a smooth two-point inverse near $(s, s)$ whose differential in local tangent-bundle coordinates at the diagonal sends basepoint/endpoint variations $(X, ℓ)$ to $ℓ - X$ . Therefore, after identifying nearby tangent spaces by a smooth local frame, $v (u, δ) : = \exp_{η (u)}^{- 1} (γ (δ)) = δ ℓ - u X + o (∣ u ∣ + δ)$ . Choose a small nonzero $a$ with $a g (ℓ, X) > 0$ and $g (ℓ - a X, ℓ - a X) = - 2 a g (ℓ, X) + a^{2} g (X, X) < 0$ . With $u = a δ$ , $v (a δ, δ) / δ \to ℓ - a X$ , so for sufficiently small $δ > 0$ the vector is timelike; its future orientation follows by continuity from the future null vector $ℓ$ for small $∣ a ∣$ . By [F14], $η (a δ) ≪ γ (δ)$ . The remaining segment $γ ∣_{[δ, λ_{p}]}$ is causal; [F15] gives $η (a δ) ≪ p$ , contradicting $p \notin I^{+} (S)$ . Hence $ℓ ⊥ T_{s} S$ .

6.1F11F15step 3.2step 4.1step 5.1step 2.2given

Return to $p \in E^{+} (S)$ and its normal generator from steps 3.2, 4.1, and 5.1, with endpoint parameter $λ_{p}$ . If a focal instant $λ_{f}$ occurred strictly before $λ_{p}$ , [F11] would give a point $r = γ (λ)$ with $λ_{f} < λ < λ_{p}$ and $r \in I^{+} (S)$ . Since $r \leq p$ , [F15] would imply $p \in I^{+} (S)$ , a contradiction. Hence $λ_{p} \leq λ_{f} \leq L$ ; for $p \in S$ the parameter is $0$ and the same bound holds.

7.1F9F11F12F16F17F18F19F22step 1.2step 6.1givenconstruct

Choose finitely many relatively compact coordinate patches covering $S$ , each contained in a patch with the two smooth normalized null normal fields from step 1.2. Their compact closures $K_{j}$ exist by choosing smaller Euclidean chart balls and applying Heine–Borel [F19] and the continuous-image theorem [F17]. For each $K_{j}$ and each branch $ℓ_{j, \pm}$ , the parameter domain $K_{j} \times [0, L]$ is compact by [F22], and the map $(s, λ) \mapsto \exp_{s} (λ ℓ_{j, \pm} (s))$ is continuous on it by [F12], [F16] and future completeness. Its image is compact by [F17]. By [F18], the finite union $C$ of these compact images is compact. Step 6.1 shows that every $p \in E^{+} (S)$ lies in $C$ .

8.1F18step 3.1step 7.1∎

By step 3.1, $E^{+} (S)$ is closed in $M$ . Since step 7.1 places it inside the compact set $C$ , it is closed in $C$ and hence compact by [F18]. This proves the claim.

Remarks

The proof does not assume that $S$ is globally achronal or that its normal bundle has a global null frame. It uses local normal branches and their positive, observer-normalized rescalings. Null Ricci nonnegativity is a direct geometric hypothesis; no Einstein equation or matter condition is inferred. A focal point is used only to bound the part of the null generator that can remain in $E^{+} (S)$ , not as a curvature singularity.

TheoremStatement: Literature-sourcedProof: AI-generatedprecheck passverified 2026-10-22 (gpt-6.1-sol)Open item page →

Penrose null geodesic incompleteness

Physical scope

A conditional Lorentzian-geometric theorem. The null Ricci inequality is assumed directly; no Einstein equation, matter-energy condition, event horizon, curvature blow-up, or singular point in the manifold is concluded.

Statement

Assume the Axiom of Choice. Let $(M, g, N)$ be a smooth time-oriented Lorentzian four-manifold of signature $(-, +, +, +)$ that is globally hyperbolic (strongly causal with compact causal diamonds). Let $Σ \subset M$ be a connected noncompact smooth spacelike Cauchy hypersurface, meaning that every inextendible causal curve meets $Σ$ exactly once. Let $S \subset M$ be a nonempty compact smooth spacelike two-surface on which both future null expansions are negative for the two local future null normal rays. Assume directly that ${Ric}_{p} (v, v) \geq 0$ for every point $p \in M$ and every null vector $v \in T_{p} M$ . Then $(M, g, N)$ is future null geodesically incomplete: at least one maximal future-directed affinely parametrized null geodesic has an affine parameter interval with finite upper endpoint. This conclusion asserts no curvature divergence or singular point belonging to $M$ .

Statement references

The Axiom of Choice

Chronology, causality, strong causality, global hyperbolicity and time functions

Spacelike hypersurfaces, acausal sets, Cauchy sets, domains of dependence and Cauchy horizons

Trapping and the asymptotic black-hole region

Future null geodesic completeness and incompleteness

Compactness of the future horismos under null completeness

A complete temporal flow identifies smooth Cauchy slices

Achronal boundaries are locally Lipschitz graphs, topological hypersurfaces and null for smooth volume densities

Temporal functions are time functions, and time functions forbid closed causal curves

Existence of a positive smooth volume density of total mass one on a smooth Lorentzian four-manifold

The Geroch volume function: a continuous strictly increasing time function whose levels are Cauchy sets

Smooth Cauchy temporal splitting: a global temporal function with Cauchy levels and an orthogonal complete flow

Invariance of domain

Facts & Assumptions

Given: Axiom of Choice; a globally hyperbolic time-oriented smooth Lorentzian four-manifold; a connected noncompact smooth spacelike Cauchy hypersurface $Σ$ ; a compact future trapped surface $S$ ; and the pointwise null Ricci inequality.

[A1]

The Axiom of Choice: AC is the assertion that every family of nonempty sets has a choice function.

[F1]

AC implies DC implies countable choice: in ZF, $A C \Rightarrow D C \Rightarrow {A C}_{ω}$ , so the Countable Choice and Dependent Choice hypotheses of the causal suppliers follow from [A1].

[F2]

Existence of a positive smooth volume density of total mass one on a smooth Lorentzian four-manifold, The Geroch volume function: a continuous strictly increasing time function whose levels are Cauchy sets, and Smooth Cauchy temporal splitting: a global temporal function with Cauchy levels and an orthogonal complete flow: under Countable Choice and Dependent Choice, a nonempty time-oriented smooth Lorentzian four-manifold has a normalized smooth volume measure; global hyperbolicity and that measure give the volume time input required by the splitting theorem; the latter then gives a smooth temporal function $T$ with Cauchy levels and a complete future timelike flow diffeomorphism $Ψ : R \times S_{0} \to M$ satisfying $T (Ψ (s, z)) = s$ .

[F3]

A complete temporal flow identifies smooth Cauchy slices: for this complete timelike product flow, projection along its flow lines identifies any smooth spacelike Cauchy hypersurface $Σ$ diffeomorphically with the slice $S_{0}$ . Thus $S_{0}$ is connected and noncompact.

[F4]

Compactness of the future horismos under null completeness: under the stated trapped-surface, null Ricci, global-hyperbolicity, and future-null-completeness assumptions, $E^{+} (S) = \partial I^{+} (S)$ is a compact topological three-manifold without boundary.

[F5]

Achronal boundaries are locally Lipschitz graphs, topological hypersurfaces and null for smooth volume densities: the boundary of an open future set is achronal; in particular no two distinct points of $E^{+} (S) = \partial I^{+} (S)$ are chronologically related.

[F6]

Temporal functions are time functions, and time functions forbid closed causal curves: a temporal function strictly increases along every nonconstant future causal curve, hence along every future timelike curve.

[F7]

Invariance of domain: assuming AC, a continuous injective map between open subsets of $R^{3}$ has open image; applying this in charts makes a continuous injective map between topological three-manifolds an open map.

[F8]

A continuous image of a compact space is compact; a continuous real-valued map on a nonempty compact space attains a maximum and a minimum; and a continuous bijection from a compact space to a Hausdorff space is a homeomorphism: a continuous real-valued map on a nonempty compact space attains its minimum, and a continuous image of a compact space is compact.

[F9]

In a Hausdorff space a point and a disjoint compact set, and two disjoint compact sets, have disjoint open neighbourhoods; hence every compact subset is closed, and in a compact Hausdorff space the compact subsets are exactly the closed ones and Separation of a topological space, connected and disconnected spaces, clopen sets, and connected subsets: compact subsets of a Hausdorff space are closed, and a nonempty subset that is both open and closed in a connected space is the whole space.

[F10]

Future null geodesic completeness and incompleteness: future null geodesic incompleteness means the existence of a maximal future null affine interval with finite upper endpoint.

[A2]

The inequality $Ric (v, v) \geq 0$ for null $v$ is a direct mathematical premise. It is not derived from a matter model, energy condition, or Einstein equation.

Proof

technique · contradiction

Given: [A1]–[A2] and [F1]–[F10]. Assume, for contradiction, that every future-directed null geodesic is future complete.

1.1A1F1assume-contragiven

Assume for contradiction that every future-directed null geodesic is future complete. By [F1], Countable Choice and Dependent Choice hold under the assumed Axiom of Choice, so the hypotheses of the volume-time, Cauchy-splitting and causal-boundary suppliers are satisfied.

2.1F2step 1.1given

Apply [F2] to obtain a volume time input, a temporal function $T$ , a smooth Cauchy slice $S_{0} = T^{- 1} (0)$ , and a complete timelike product flow $Ψ$ . The splitting satisfies $T (Ψ (s, z)) = s$ . Hence $T$ tends to $+ \infty$ and $- \infty$ along the two ends of every orbit; continuity prevents an endpoint in $M$ , so each orbit is an inextendible timelike curve.

3.1F3step 2.1given

Apply [F3] to identify the given hypersurface $Σ$ diffeomorphically with $S_{0}$ by projection along these flow lines. Therefore $S_{0}$ is connected and noncompact.

3.2F2F4F6F8step 2.1givencontradiction

The restriction $T ∣_{S}$ is continuous and $S$ is nonempty compact, so [F8] gives a point $s_{0} \in S$ at which $T ∣_{S}$ is minimal. If $s_{0} \in I^{+} (S)$ , some $s_{1} \in S$ satisfies $s_{1} ≪ s_{0}$ ; since $T$ is temporal, [F6] gives $T (s_{1}) < T (s_{0})$ , contradicting minimality. Thus $s_{0} \notin I^{+} (S)$ . As $s_{0} \in S \subseteq J^{+} (S)$ by reflexivity, $s_{0} \in E^{+} (S)$ , so this boundary is nonempty.

3.3F2F5step 2.1givenconstruct

Define $π : M \to S_{0}$ by $π (Ψ (s, z)) = z$ . This is continuous because $Ψ^{- 1}$ is smooth. If $x, y \in E^{+} (S)$ have $π (x) = π (y)$ and $x \neq y$ , their flow coordinates differ, so the timelike orbit between them makes them chronologically related. This contradicts the achronality in [F5]. Hence $π ∣_{E^{+} (S)}$ is injective.

4.1F2F4F7step 3.3construct

The domain $E^{+} (S)$ is a topological three-manifold without boundary by [F4], and $S_{0}$ is a smooth three-manifold by [F2]. For each $x \in E^{+} (S)$ , choose manifold charts about $x$ and $π (x)$ and shrink the source chart so its image lies in the target chart. In these coordinates $π$ is continuous and injective between open subsets of $R^{3}$ ; [F7] makes its image open. Taking the union over $x$ shows $π (E^{+} (S))$ is open in $S_{0}$ .

4.2F2F4F8F9step 3.3

By [F4], $E^{+} (S)$ is compact, and $π$ is continuous, so [F8] makes $π (E^{+} (S))$ compact. The smooth manifold $S_{0}$ is Hausdorff; [F9] therefore makes this image closed in $S_{0}$ .

5.1F9F10step 3.1step 3.2step 4.1step 4.2contradictiondischarge-contradiction∎

The image $π (E^{+} (S))$ is nonempty by step 3.2 and both open and closed by steps 4.1 and 4.2. Since $S_{0}$ is connected by step 3.1, [F9] implies $π (E^{+} (S)) = S_{0}$ . But $π (E^{+} (S))$ is compact by step 4.2, which would make $S_{0}$ compact, contradicting step 3.1. Thus the assumed future null completeness is false. By [F10], at least one maximal future-directed null geodesic has finite upper affine endpoint, which is the asserted conclusion.

Remarks

The null Ricci condition is a geometric hypothesis in the library curvature convention; it is not silently replaced by an Einstein equation or a matter energy condition. The conclusion is affine incompleteness only. It neither identifies an incomplete endpoint inside $M$ nor asserts curvature blow-up, cosmic censorship, or the existence of an event horizon. A trapped surface is not itself an event horizon; that comparison requires separate asymptotic and global hypotheses.

5 · Examples, experiments and counterexamples

None yet.

Trapped surfaces and Penrose incompleteness

1 · Prerequisites

2 · Summary

3 · Logical flowchart

4 · Definitions, postulates, theorems and proofs

Future null geodesic completeness and incompleteness

Definition

Definition references

Conventions and scope

A complete temporal flow identifies smooth Cauchy slices

Statement

Statement references

Facts & Assumptions

Proof

Remarks

Trapping and the asymptotic black-hole region

Definition

Definition references

Conventions and scope

Compactness of the future horismos under null completeness

Statement

Statement references

Facts & Assumptions

Proof

Remarks

Penrose null geodesic incompleteness

Statement

Statement references

Facts & Assumptions

Proof

Remarks

5 · Examples, experiments and counterexamples

Sources