Alphabeta Math
LemmaStatement: AI-adaptedProof: AI-generatedprecheck passjudge pass (z-ai/glm-5.2)audited 2026-07-27
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.

Complete metrizability: admitting a topologically equivalent complete metric is preserved by homeomorphism and by closed subspaces, and (0,∞) has it without being complete

Statement

Let (X,d) be a metric space (Metric space: d(x,y)=0 iff x=y, symmetry, and the triangle inequality; pseudometric and ultrametric) and let Td be its metric topology (The metric topology: a set is open when every one of its points has a ball around it inside the set; closed means open complement). Call Td completely metrizable if some metric ρ on X is topologically equivalent to d, that is Tρ=Td (Topologically, uniformly and Lipschitz equivalent metrics on a set), and makes (X,ρ) complete (Complete metric space: every Cauchy sequence converges in the space). Then:

  1. Homeomorphism invariance. Let (Y,e) be a metric space and let h:X→Y be a bijection (Injection, surjection, bijection) such that h and h−1 are continuous (Continuity of a map between metric spaces, at a point and globally, in the ε-δ form). If Td is completely metrizable then so is Te.
  2. Closed subspaces. If Td is completely metrizable and A⊆X is closed in (X,d), then TdA is completely metrizable, dA being the subspace metric (Isometry, isometric embedding, and the subspace metric on a subset).
  3. The property is strictly weaker than completeness. Let P:=(0,∞)⊆R (Intervals of R: the nine order-convex forms, nondegeneracy, and length) carry d(x,y):=∣x−y∣ (The absolute value makes R a metric space: d(x,y)=∣x−y∣ is a metric, its open balls are the intervals (x−r,x+r), and it is unbounded). Then (P,d) is not complete, while ρP(x,y)  :=  ∣x−y∣  +  ∣1x−1y∣ is a complete metric on P with TρP=Td. So Td is completely metrizable although no completeness assumption holds for d itself.

Complete metrizability is a condition on the collection of open sets alone: the metric is quantified over and does not survive into the statement. That is exactly what completeness fails to be, and claim 3 shows the two conditions are genuinely different rather than merely stated differently.

Facts & Assumptions

Given: A metric space (X,d); a metric space (Y,e) and a bijection h:X→Y with h and h−1 continuous; a subset A⊆X closed in (X,d) and carrying the subspace metric dA; the set P:=(0,∞) with d(x,y)=∣x−y∣; a real ε>0.

[L2]

The subspace metric is the restriction, dA=d↾(A×A); so a sequence in A is dA-Cauchy exactly when it is d-Cauchy, and converges to a∈A in (A,dA) exactly when it converges to a in (X,d) (Isometry, isometric embedding, and the subspace metric on a subset, Cauchy sequence in a metric space, Convergence of a sequence in a metric space: xk→x iff d(xk,x)→0 in R).

[L3]

A complete subspace of any metric space is closed, and a closed subspace of a complete space is complete (Closed subspaces of complete metric spaces are complete; the converse under countable choice).

[L4]

An isometric embedding f satisfies dY(f(u),f(v))=dX(u,v), and a subset U of its source is open exactly when f[U] is open in the image with its subspace metric (Isometry, isometric embedding, and the subspace metric on a subset, An isometric embedding is injective and carries the metric topology of the source onto the subspace topology of its image).

[L8]
[L9]

For x>0 the reciprocal 1/x is positive, 1/x=1/y forces x=y, and y>a/2>0 gives 1/(ay)<2/a2 (Inverses of positives are positive, and reciprocation reverses order, Reciprocals and order: 1/r against 1).

[L10]

A bijection h satisfies h[U]=(h−1)−1[U] and h−1[h[U]]=U for every U in its source (Injection, surjection, bijection).

Proof

technique · direct
1.1

If σ and σ′ are metrics on one set Z, then Tσ=Tσ′ holds exactly when both of the following do: for every z∈Z and real ε>0 there is a real δ>0 with σ′(z,w)<ε whenever σ(z,w)<δ, and the same with σ and σ′ interchanged. Indeed the two conditions say that the two identity maps are ε-δ continuous, which by [L1] says that each topology is contained in the other.

L1
1.2

For claim 1 let ρ be as in [A1] and put ρ′(y,y′):=ρ(h−1(y),h−1(y′)) for y,y′∈Y. This is a metric on Y: (M2) and (M3) are inherited pointwise from ρ, and (M1) holds because h−1 is injective, so ρ′(y,y′)=0 gives h−1(y)=h−1(y′) and hence y=y′. By construction ρ′(h(x),h(x′))=ρ(x,x′), so h:(X,ρ)→(Y,ρ′) is a bijective isometric embedding.

A1L5L10construct
1.3

For claim 3 put σ(x,y):=∣1/x−1/y∣ for x,y∈P. This is a metric on P: it is nonnegative and symmetric, it satisfies the triangle inequality because the absolute value does, and σ(x,y)=0 gives 1/x=1/y and hence x=y.

L5L9construct
1.4

A sum of two metrics on one set is again a metric, since symmetry and the triangle inequality add, the sum of two nonnegative reals is nonnegative, and the sum vanishes exactly when both summands do. Hence ρP=d+σ is a metric on P, and d(x,y)≤ρP(x,y) and σ(x,y)≤ρP(x,y) for all x,y∈P.

L5
1.5

Let a∈P and let ε>0 be real; put δ:=min⁡{ a/2, ε/(1+2/a2) }, a positive real. For y∈P with ∣a−y∣<δ one has y>a−a/2=a/2, hence ay>a2/2 and σ(a,y)=∣a−y∣/(ay)<2∣a−y∣/a2, so ρP(a,y)<∣a−y∣⋅(1+2/a2)<ε.

L5L9algebra
1.6

The sequence xk:=1/(k+2) has all its terms in P and converges in R to 0, which is not in P; so P is not sequentially closed in R and therefore not closed in (R,∣x−y∣).

L6L7L8L9
1.7

For claim 2 let ρ be as in [A1]. Since Tρ=Td and A is closed in (X,d), the set A is closed in (X,ρ) as well, so (A,ρA) is complete by [L3], ρA being the restriction of ρ to A×A.

A1L2L3
2.1

Claim 1, completeness: let (yk) be a ρ′-Cauchy sequence in Y. By step 1.2 the sequence (h−1(yk)) is ρ-Cauchy, so by [A1] it converges in (X,ρ) to some x∈X, and then ρ′(yk,h(x))=ρ(h−1(yk),x)→0, that is yk→h(x) in (Y,ρ′). So (Y,ρ′) is complete.

step 1.2A1L5
2.2

Claim 1, topology: by [L4] applied to the bijective isometric embedding h of step 1.2, whose image is all of Y with ρ′ itself as subspace metric, a set U⊆X is ρ-open exactly when h[U] is ρ′-open. And U is d-open exactly when h[U] is e-open, since h[U]=(h−1)−1[U] is e-open for d-open U by continuity of h−1, and conversely U=h−1[h[U]] is d-open for e-open h[U] by continuity of h. As Tρ=Td by [A1], the two equivalences give Tρ′=Te.

step 1.2A1L1L4L10
2.3

Claim 2, topology: apply step 1.1 to ρ and d on X, which is legitimate by [A1], and restrict the two resulting ε-δ conditions to points of A; since ρA and dA are the restrictions of ρ and d, the same δs witness the two conditions of step 1.1 for ρA and dA on A, whence TρA=TdA.

step 1.1step 1.7A1L2
2.4

Claim 3, topology: by step 1.4 the identity (P,ρP)→(P,d) satisfies the first condition of step 1.1 with δ:=ε, and by step 1.5 the identity (P,d)→(P,ρP) satisfies the second; so TρP=Td.

step 1.1step 1.4step 1.5
2.5

Claim 3, failure of completeness for d: were (P,d) complete, [L3] would make P closed in (R,∣x−y∣), contradicting step 1.6. So (P,d) is not complete.

step 1.6L2L3L6
2.6

Claim 3, completeness of ρP: let (xk) be a ρP-Cauchy sequence in P. By the two inequalities of step 1.4 both (xk) and (1/xk) are Cauchy sequences of reals, so by [L6] they converge, say xk→L and 1/xk→c; and L≥0 and c≥0, all terms being positive.

step 1.3step 1.4L2L6L7
3.1

Claim 1 is established: ρ′ is a complete metric on Y with Tρ′=Te, so Te is completely metrizable.

step 2.1step 2.2
3.2

Claim 2 is established: ρA is a complete metric on A with TρA=TdA, so TdA is completely metrizable.

step 1.7step 2.3
3.3

Continuing step 2.6: xk⋅(1/xk)=1 for every k, so Lc=1 by multiplicativity of limits; hence L≠0, so L>0 and L∈P, and c=1/L.

step 2.6L7L9
4.1

Hence ρP(xk,L)=∣xk−L∣+∣1/xk−1/L∣→0 by additivity of limits, that is xk→L in (P,ρP) with L∈P; every ρP-Cauchy sequence in P therefore converges in P, and (P,ρP) is complete.

step 2.6step 3.3L7
5.1

Claim 3 is established by step 2.4, step 2.5 and step 4.1, and claims 1 and 2 by step 3.1 and step 3.2.

step 2.4step 2.5step 3.1step 3.2step 4.1∎

Remarks

  • What claim 3 decides, and what it leaves open. It settles that "carries a complete metric" is strictly weaker than "this metric is complete", on the cheapest example available here. It does not characterise the topologies that are completely metrizable. The classical characterisation is Alexandroff's theorem — a subspace of a complete metric space is completely metrizable exactly when it is a Gδ subset — and it is out of reach at this point in the library, needing countable intersections of open sets, the Baire category theorem, and a metric built as a convergent series of terms ∣1/dist⁡(x,X∖Un)−1/dist⁡(y,X∖Un)∣. None of that is available here.

  • The one-sided reading of claim 2. A closed subspace of a completely metrizable space is completely metrizable. An open one is too, and so is any countable intersection of open sets, but that is Alexandroff's theorem and is not proved here, so nothing on this page licenses either. Nor does anything here decide a subspace that is neither open nor closed: (0,1] inside R is completely metrizable and Q is not, and both facts need machinery this page does not have.

  • Where the term is fixed. This item introduces "completely metrizable" as a property of a metric topology, since a topology here is a collection of subsets (The metric topology: a set is open when every one of its points has a ball around it inside the set; closed means open complement) rather than an abstract space. A later page of this library restates it for a general topological space; that restatement is a transfer of this definition along the identification of the two developments, not a second notion.

  • Claim 1 is what makes the property topological at all. Read literally, the definition already refers to Td alone, so the content of claim 1 is that the property travels between different underlying sets: a homeomorphism transports one complete metric to another, by making itself an isometry (An isometric embedding is injective and carries the metric topology of the source onto the subspace topology of its image). Completeness itself does not travel that way, since a homeomorphism need not be an isometry for the given metrics, and that is the whole difference.

Depends on

Used by

Dependency tree · two levels

66 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