Alphabeta Math
TheoremStatement: Literature-sourcedProof: AI-adaptedPipeline-generatedprecheck passaudited 2026-09-30
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.

The cut locus of a point is closed

Statement

Assume exactly the inherited Axiom of Countable Choice ACω, carried by the declared exponential-domain, cut-time and sequential-closure suppliers. Let (M,g) be a complete, connected, boundaryless, finite-dimensional Riemannian manifold, let p∈M, and let Cut⁡(p)={exp⁡p(cp(v)v):v∈SpM, cp(v)<+∞} be the cut locus of p from Cut point and cut locus of a point. Then Cut⁡(p) is closed in the metric space (M,dg).

In dimension zero SpM=∅, so Cut⁡(p)=∅, which is closed. No compactness of M is assumed; the only compactness used is the sequential compactness of the finite-dimensional unit sphere SpM.

Facts & Assumptions

Given: The complete connected boundaryless finite-dimensional Riemannian manifold (M,g), the point p∈M, the cut time c=cp:SpM→(0,+∞], the unit sphere SpM and the cut locus Cut⁡(p).

[A1]

The choice assumption is ACω of The Axiom of Countable Choice (ACω), inherited through the declared suppliers and spent in this proof exactly at the two points flagged in steps 1.1 and 5.1; no full Axiom of Choice and no dependent choice is used.

[F1]

The cut locus is Cut⁡(p)={exp⁡p(cp(v)v):v∈SpM, cp(v)<+∞}, and along the unit-speed radial geodesic in direction v one has γv(t)=exp⁡p(tv) (Cut point and cut locus of a point).

[F2]

If the cut time cp(v) is finite then cp(v)∈Ap(v)={t≥0:dg(p,γ(t))=t}, that is dg(p,γv(cp(v)))=cp(v) (Minimizing along a geodesic is an initial interval property).

[F3]

dg is a finite metric on the connected Riemannian manifold M (Riemannian distance is a metric), so it satisfies (M1) separation, (M2) symmetry and (M3) the triangle inequality (Metric space: d(x,y)=0 iff x=y, symmetry, and the triangle inequality; pseudometric and ultrametric); convergence xk→x in (M,dg) means dg(xk,x)→0 in R (Convergence of a sequence in a metric space: xk→x iff d(xk,x)→0 in R).

[F4]

SpM is sequentially compact: every sequence of unit tangent vectors at p has a subsequence converging in the norm metric of TpM to a unit tangent vector at p (Unit spheres in finite-dimensional normed spaces are sequentially compact).

[F5]

The cut time is positive at every unit vector, and it is continuous in the extended sense: whenever vk→v in SpM, if c(v)<+∞ then c(vk)→c(v), while if c(v)=+∞ then for every M0>0 there is k0 with c(vk)>M0 for all k≥k0 (Cut time is positive and continuous).

[F6]

On the complete manifold (M,g) the Hopf–Rinow equivalent condition 3 holds: for every point of M the fibre exponential domain is all of the tangent space, so Ep=TpM (Hopf–Rinow theorem).

[F7]

The exponential map is smooth on its domain, hence continuous; consequently each fibre restriction exp⁡p is smooth on the open set Ep (The exponential domain is open and the exponential map is smooth). The target manifold topology is the dg topology (The riemannian distance topology is the manifold topology), so this continuity also gives convergence in (M,dg).

[F8]

On TpM the Riemannian metric is a positive definite symmetric bilinear form (Riemannian metric and riemannian manifold), the pointwise norm is ∣v∣g=g(v,v) (Pointwise norm and angle from a riemannian metric), and the induced inner-product norm satisfies ∣λv∣g=∣λ∣ ∣v∣g and ∣u+v∣g≤∣u∣g+∣v∣g (The induced length is a norm).

[F9]

In a metric space a subset F is closed if and only if it is sequentially closed, that is, if and only if every sequence in F that converges in the ambient space has its limit in F (A point lies in the closure of A iff some sequence in A converges to it, and a set is closed iff it is sequentially closed).

Proof

technique · sequential closedness, read through the metric-space equivalence between closed and sequentially closed sets: a convergent sequence of cut points has cut times equal to the distances to $p$, hence bounded; compactness of the unit sphere produces a limit direction; continuity of the cut time makes the limiting cut time finite; and continuity of the exponential map realises the limit as the corresponding cut point
1.1A1F1F2F3given

A convergent sequence of cut points and its cut times. [A1, F1, F2, F3, given] Let (qk) be a sequence in Cut⁡(p) with qk→q in (M,dg). For every k the set {v∈SpM:cp(v)<+∞, exp⁡p(cp(v)v)=qk} is nonempty by [F1], and selecting one vk for each k∈N is exactly the countable choice permitted by [A1]; then c(vk)<+∞ and qk=exp⁡p(c(vk)vk) for every k. By [F2] a finite cut time is attained, c(vk)∈Ap(vk), so c(vk)=dg(p,γvk(c(vk)))=dg(p,exp⁡p(c(vk)vk))=dg(p,qk), using γv(t)=exp⁡p(tv) [F1]. Since dg is a metric [F3], (M2) and (M3) give ∣dg(p,qk)−dg(p,q)∣≤dg(qk,q) for every k, and dg(qk,q)→0 because qk→q [F3]; hence c(vk)→dg(p,q)=:L with L∈[0,+∞), so the real sequence (c(vk)) is bounded.

2.1F3F4givenstep 1.1

A convergent subsequence of directions. [F3, F4, given, step 1.1] Since (c(vk)) is bounded (step 1.1) and SpM is sequentially compact [F4], there is a subsequence vkj→v with v∈SpM. A subsequence of the convergent sequence (c(vk)) has the same limit L [F3, step 1.1], so c(vkj)→L. Relabelling that subsequence, assume from now on that vk→v in SpM and c(vk)→L.

3.1F3F5givenstep 2.1

The limiting cut time is finite. [F3, F5, given, step 2.1] Suppose c(v)=+∞. By the infinite-value clause of the continuity of the cut time [F5], for every M0>0 there is k0 with c(vk)>M0 for all k≥k0; taking M0=L+1 contradicts c(vk)→L (step 2.1) [F3]. Hence c(v)<+∞, and the finite-value clause of [F5] gives c(vk)→c(v). The two limits c(v) and L of the same sequence agree: for every ε>0 both ∣c(vk)−c(v)∣<ε and ∣c(vk)−L∣<ε hold for all large k, whence ∣c(v)−L∣<2ε, and ε>0 was arbitrary. Therefore c(v)=L<+∞.

4.1F1F6F7F8givenstep 1.1step 2.1step 3.1

The limit point is the corresponding cut point. [F1, F6, F7, F8, given, step 1.1, step 2.1, step 3.1] Put wk:=c(vk)vk and w:=c(v)v. The vectors wk lie in Ep because exp⁡p(wk)=qk [F1], and by [F6] Ep=TpM, so also w∈Ep. For the convergence wk→w, the norm estimates of [F8] give, since ∣vk∣g=∣v∣g=1 and c(vk)>0, ∣wk−w∣g=∣c(vk)vk−c(vk)v+c(vk)v−c(v)v∣g≤c(vk) ∣vk−v∣g+∣c(vk)−c(v)∣ ∣v∣g=c(vk)∣vk−v∣g+∣c(vk)−c(v)∣, and the right-hand side tends to 0 because c(vk) is bounded by step 1.1, vk→v (step 2.1) and c(vk)→c(v) (step 3.1) [F8]. As exp⁡p is smooth, hence continuous, on Ep [F7], exp⁡p(wk)→exp⁡p(w); but exp⁡p(wk)=qk→q (steps 1.1 and 2.1), so q=exp⁡p(w)=exp⁡p(c(v)v)=γv(c(v)) by γv(t)=exp⁡p(tv) [F1].

5.1F1F9givenstep 4.1

Conclusion: the cut locus is closed. [F1, F9, given, step 4.1] Step 4.1 exhibits the limit q as exp⁡p(c(v)v) with v∈SpM a unit vector and c(v)=L<+∞; therefore q∈Cut⁡(p) by the definition of the cut locus [F1]. Since the convergent sequence (qk) in Cut⁡(p) was arbitrary, Cut⁡(p) is sequentially closed, and by the metric-space equivalence between closedness and sequential closedness [F9] it is closed in (M,dg).

6.1A1F1F4F5F6F9givenstep 1.1step 3.1step 4.1step 5.1

Boundary and choice audit. [A1, F1, F4, F5, F6, F9, given, step 1.1, step 3.1, step 4.1, step 5.1] In dimension zero TpM={0p}, so ∣0p∣g=0≠1 and SpM=∅; hence Cut⁡(p)=∅ by [F1], and the empty set is closed in (M,dg) [F9], both because its complement M is open and because the sequential criterion holds vacuously. The empty manifold has no point p. In dimension one the unit sphere has exactly two points and every step applies verbatim, the sequential compactness [F4] being immediate for a two-point set. Since c takes values in (0,+∞] [F5], the limiting time c(v)=L of step 3.1 is positive, so the degenerate value L=0 (which would force the cut point to be the base point p) does not occur, and all directions occurring above are unit vectors, so no constant geodesic arises. Completeness enters only through the global exponential domain [F6] referenced in step 4.1 and through the cut-time continuity theorem [F5]; no compactness of M is assumed, the compactness used being that of the finite-dimensional sphere SpM [F4]. The choice audit: ACω of [A1] is spent exactly twice, once in step 1.1, where one direction vk is selected from each of the countably many nonempty fibres over the qk, and once in step 5.1 through the direction of [F9] that converts sequential closedness into closedness (contrapositively, one point is selected from each of the countably many nonempty sets B(q,1/(n+1))∩Cut⁡(p)).

□

Source locator

Lee, Riemannian Manifolds: An Introduction to Curvature, Chapter 10, printed pp.173-190, develops the cut locus of a point; Datar, Lectures on Riemannian Geometry, Lecture 23, sections 23.2-23.3, printed pp.163-172, presents the cut locus and the cut time. The closedness of Cut⁡(p) is proved above as sequential closedness, from the continuity of the cut time and the compactness of the finite-dimensional unit sphere; no source text is quoted.

Depends on

Used by

Dependency tree · two levels

90 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.

Sources