Alphabeta Math
TheoremStatement: AI-adaptedProof: AI-generatedprecheck pass
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.

Two-endpoint limit curve theorem: bounded lengths give joining limits, unbounded lengths give two inextendible tails with the closure relation

Statement

Assume the Axioms of Countable Choice and Dependent Choice (The Axiom of Countable Choice (ACω), The axiom of dependent choice: a relation in which every element is related to something admits an N-indexed chain). Let (M,g,N) be a time-oriented smooth Lorentzian four-manifold and let h denote the complete conformally rescaled metric h′ constructed in The auxiliary Riemannian metric h=g+2n-flat tensor n-flat is complete after a proper-exhaustion conformal factor, and its closed bounded sets are compact clause (b). All distances here are within a connected component; each component is complete and its closed dh-bounded sets are compact by clauses (c) and (d). Endpoint convergence puts the joining curves in the component of x for all sufficiently large k, since components of a manifold are open. Let αk:[0,bk]→M,0<bk<∞, be future causal curves (Causal curves, chronological and causal reachability, and inextendibility) parametrized by h-arclength, ℓh(αk∣[s,t])=t−s for 0≤s≤t≤bk, with αk(0)→x∈M and αk(bk)→z∈M. Passing to a subsequence, let b:=lim⁡kbk∈[0,∞]. Then:

(i) Bounded case. If 0<b<∞, there is a subsequence and a future causal curve α:[0,b]→M with α(0)=x, α(b)=z and dh(α(s),α(t))≤t−s for s≤t, such that the subsequence converges to α uniformly in dh on [0,b] after the following common-domain reparametrization when b>0: α~k(t):=αk ⁣(bkbt),0≤t≤b. Thus α~k(0)=αk(0) and α~k(b)=αk(bk). If b=0, then x=z and the curves converge to the constant map at x uniformly on their varying domains in the sense that sup⁡0≤t≤bkdh(αk(t),x)→0. The constant map is the continuous limit, not a future causal curve in the nonzero-velocity curve class.

(ii) Unbounded case. If b=∞, there are a future-inextendible future causal curve α+:[0,∞)→M with α+(0)=x and a past-inextendible future causal curve α−:(−∞,0]→M with α−(0)=z, such that a subsequence converges to α+ uniformly on every compact subinterval of [0,∞) and to α− uniformly on every compact subinterval of (−∞,0] (where α− is the limit of the curves s↦αk(bk+s), s≤0, for large k). Moreover, for all t+≥0 and t−≤0 there are event pairs uk,vk∈M with uk≤vk, uk→α+(t+) and vk→α−(t−). Thus (α+(t+),α−(t−)) lies in the sequential closure of the causal relation, meaning precisely that it is a limit of pairs in the relation ≤.

(iii) Coincident endpoints. If x=z, then in case (i) with b>0 the curve α is a nonconstant closed future causal curve through x (α(0)=α(b)=x), and in case (ii) the concatenation γ of α− and α+ is an inextendible future causal curve γ:R→M with γ(0)=x.

(iv) Imprisonment. Let (M,g,N) be strongly causal (Chronology, causality, strong causality, global hyperbolicity and time functions clause (b)) and suppose all images αk([0,bk]) are contained in one compact set K. Then b≤L(K)<∞ where L(K) is the length bound of Strong causality bounds the auxiliary length of causal curves contained in a compact set; in particular the case b=∞ is excluded.

Facts & Assumptions

Given: Countable and dependent choice, the Lorentzian manifold, the auxiliary metric h, and the arclength-parametrized sequence αk with its endpoint limits x,z and the value b=lim⁡kbk.

[L1]

The basic limit curve lemma: for a sequence of future causal h-arclength-parametrized curves defined on intervals containing [0,T] with initial points converging, a subsequence converges uniformly in dh on [0,T] to a future causal curve satisfying (a1)-(a4) there; for intervals eventually containing [0,n] for every n, a diagonal subsequence converges on every compact subinterval of [0,∞) and the limit is future-inextendible; the same holds in the time-reversed orientation for the past side (Basic limit curve lemma: arclength-parametrized causal curves have causal limit curves on compact parameter intervals and future-inextendible limits on half-lines clauses (a) and (b)).

[L2]

For s<t inside the domain of a future causal curve β, the restriction β∣[s,t] is a future causal curve from β(s) to β(t); hence β(s)≤β(t) (Chart independence, concatenation, reparametrization and transitivity for causal curves clause (a), Causal curves, chronological and causal reachability, and inextendibility clause (c)).

[L3]

Concatenation of two future causal curves at a common endpoint is a future causal curve; a curve that is both future- and past-inextendible is inextendible (Chart independence, concatenation, reparametrization and transitivity for causal curves clause (d), Causal curves, chronological and causal reachability, and inextendibility clause (d), Equivalence of the absolutely-continuous and continuous causal conventions, and one-sided inextendible extensions clause (d)).

[L4]

On each connected component dh is a complete metric with compact closed bounded balls by The auxiliary Riemannian metric h=g+2n-flat tensor n-flat is complete after a proper-exhaustion conformal factor, and its closed bounded sets are compact clauses (c) and (d), inducing the manifold topology; for every locally absolutely continuous β and s<t, dh(β(s),β(t))≤ℓh(β∣[s,t]). In particular, h-arclength parametrization gives the upper bound dh(β(s),β(t))≤t−s, not equality in general (Riemannian distance on a connected manifold, Riemannian distance is a metric, The riemannian distance topology is the manifold topology).

[L5]

In a strongly causal spacetime, causal curves with image in the compact set K have h-length at most L(K), and h-arclength-parametrized curves with image in K have domain length at most L(K) (Strong causality bounds the auxiliary length of causal curves contained in a compact set).

[L6]

A uniformly Lipschitz sequence of maps from a compact interval into a compact metric space has a uniformly convergent subsequence, and its limit has the same Lipschitz bound (A pointwise convergent subsequence of a uniformly Lipschitz family into a compact metric space converges uniformly on compact subintervals).

[L7]

In an adapted chart from Causal cone domination in adapted charts and Lipschitz-graph bounds for causal curves, future causal curves satisfy a uniform cone-speed estimate and the time coordinate is strictly increasing. On a compact coordinate ball, smoothness of h and the local metric comparison with the Euclidean metric give constants c,C0>0 such that c∣x˙∣e2≤h(γ˙,γ˙)≤C0∣x˙∣e2. Thus the h-length of a causal segment contained in a smaller ball is at most C Δx0, using the cone-speed estimate. To compare distance and coordinates, choose concentric coordinate balls Br⊂BR with compact closure in the chart and 2r≤R−r. For y,z∈Br, every path staying in BR has length at least c ∣x(y)−x(z)∣e; every path exiting BR has initial length at least c(R−r)≥c ∣x(y)−x(z)∣e before its first exit. Taking the infimum over paths gives ∣x(y)−x(z)∣e≤c−1/2dh(y,z). Hence a dh-Lipschitz curve is locally Lipschitz in chart coordinates (Causal cone domination in adapted charts and Lipschitz-graph bounds for causal curves clauses (a)-(c), The auxiliary Riemannian metric h=g+2n-flat tensor n-flat is complete after a proper-exhaustion conformal factor, and its closed bounded sets are compact, Local comparison of a riemannian metric with the euclidean metric).

Proof

technique · case-analysis
1.1L4L6givenconstruct

(Common-domain reparametrization in the bounded positive case.) Suppose 0<b<∞. Define α~k(t):=αk ⁣(bkbt),0≤t≤b. Since bk/b→1, after discarding finitely many terms these curves are 2-Lipschitz for dh: the arclength parametrization gives dh(α~k(s),α~k(t))≤ℓh(α~k∣[s,t])=bkb∣t−s∣≤2∣t−s∣. Their initial points converge to x, and their images lie in one closed h-ball about x of radius 2b+1, which is compact by [L4]. Applying [L6] gives a subsequence converging uniformly on [0,b] to a 2-Lipschitz map α. The sharper inequality dh(α~k(s),α~k(t))≤bkb∣t−s∣ and bk/b→1 imply dh(α(s),α(t))≤∣t−s∣. Uniform convergence at the endpoints gives α(0)=x and α(b)=z.

1.2L1L4L6L7given

(The limit is future causal.) Let [s,t] be a parameter interval such that α([s,t]) lies in a strictly convex normal set C. Since this image is compact and C is open, uniform convergence implies α~k([s,t])⊂C for all sufficiently large k. By Equivalence of the absolutely-continuous and continuous causal conventions, and one-sided inextendible extensions clause (a), each approximating causal subarc has a future-causal position vector wk=exp⁡α~k(s)−1(α~k(t)). Smoothness of the two-point inverse on C×C gives wk→w=exp⁡α(s)−1(α(t)); closedness of the future cone gives that w is future causal or zero. Thus α is geodesically connectible. It is locally Lipschitz in charts by its dh-Lipschitz bound and [L7], hence locally absolutely continuous. At almost every differentiability parameter t<b, choose a relatively compact adapted chart U around α(t) and h0>0 so α([t,t+h])⊂U for 0<h≤h0. For all sufficiently large k the approximating arcs also lie in U and have h-length (bk/b)h. By [L7], bkbh≤C(x0(α~k(t+h))−x0(α~k(t))) for a chart constant C. Passing to the limit and using bk/b→1 gives x0(α(t+h))−x0(α(t))≥h/C. At differentiability points this yields (x0∘α)′(t)≥1/C>0. For small h>0 the position vector exp⁡α(t)−1(α(t+h)) is future causal or zero by the preceding connectibility argument; divided by h it tends in coordinates to α′(t), because the differential of the second-variable inverse at the diagonal is the identity (The joint exponential map: smoothness, invertible differential at the diagonal, and the small-velocity two-point inverse clauses (b),(c)). The future cone is closed and α′ has positive time component, so α′(t) is nonzero future causal almost everywhere. This proves α is a future causal curve and is not constant on any nondegenerate interval.

1.3L1givenchoose

(The future tail in the unbounded case.) Suppose b=∞. For each fixed T>0, [0,T] is contained in [0,bk] for all large k. Applying clause (b) of [L1] by diagonal extraction over integer T gives a subsequence converging uniformly on every compact subinterval of [0,∞) to a future-inextendible future causal curve α+:[0,∞)→M with α+(0)=x.

1.4L1givenchoose

(The past tail in the unbounded case.) Define δk(s):=αk(bk+s) on [−bk,0]. For each fixed T>0, [−T,0] lies in [−bk,0] for all large k. Apply the time-reversed half-line result of [L1] and diagonalize to obtain a past-inextendible future causal curve α−:(−∞,0]→M with α−(0)=z, such that δk converges uniformly to α− on compact subintervals. A nested subsequence retains both tail limits.

1.5L4givenalgebra

(The zero-length case.) If b=0, then dh(x,z)≤dh(x,αk(0))+dh(αk(0),αk(bk))+dh(αk(bk),z)≤o(1)+bk+o(1), so x=z. For every t∈[0,bk], dh(αk(t),x)≤dh(αk(0),x)+t≤o(1)+bk. Hence the curves converge uniformly on their varying domains to the constant curve at x.

1.6L5givenalgebra

(Imprisonment.) If the spacetime is strongly causal and all images lie in a compact K, [L5] gives bk≤L(K) for every k. Taking limits yields b≤L(K)<∞, excluding the unbounded case.

2.1step 1.3step 1.4L2given

(The closure relation.) Fix t+≥0 and t−≤0. For all sufficiently large k, t+≤bk+t−, so by [L2], αk(t+)≤αk(bk+t−)=δk(t−). The uniform-on-compact limits in steps 1.3-1.4 give endpoints converging respectively to α+(t+) and α−(t−). This is the stated sequential closure relation.

2.2step 1.1step 1.2step 1.3step 1.4L3given

(Coincident endpoints.) If x=z and 0<b<∞, step 1.2 gives a future causal curve α from x to x which is nonconstant on every nondegenerate interval. If x=z and b=∞, concatenate α− and α+ at their common endpoint x. The concatenation is future causal by [L3], and is past- and future-inextendible because its respective tails are; hence it is inextendible.

3.1step 1.1step 1.2step 1.3step 1.4step 2.1step 1.5step 1.6step 2.2∎

(Conclusion.) Clause (i) for b>0 is steps 1.1-1.2 and for b=0 is step 1.5; clause (ii) is steps 1.3, 1.4 and 2.1; clause (iii) is step 2.2; and clause (iv) is step 1.6.

Remarks

  • The two alternatives are exclusive in the applications. The bounded alternative produces a joining curve between the limiting endpoints; the unbounded alternative produces two one-sided curves related through the limit of the sequence, not a joining curve. The closure relation is recorded in the weak sequential form actually used downstream.

  • Role of the choice assumptions. Countable choice is used for the diagonal subsequence selections, dependent choice is inherited from the manifold and curve calculus suppliers; no arbitrary family of curves is selected.

  • No length semicontinuity. The theorem does not compare Lorentzian lengths; that is Lorentzian length as a refined sum of local separations, and upper semicontinuity of the length under uniform convergence.

Depends on

Dependency tree · 2 levels

Sources