Alphabeta Math
TheoremStatement: AI-adaptedProof: AI-generatedSession-authored (Fable 5 assisted)precheck 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.

A sequentially compact metric space is totally bounded, proved from the axiom of dependent choice

Statement

Assume the Axiom of Dependent Choice (The axiom of dependent choice: a relation in which every element is related to something admits an N\mathbb{N}-indexed chain). Let (X,d)(X,d) be a sequentially compact metric space (Countably compact, sequentially compact and limit point compact metric spaces, Metric space: d(x,y)=0d(x,y) = 0 iff x=yx = y, symmetry, and the triangle inequality; pseudometric and ultrametric). Then (X,d)(X,d) is totally bounded (Finite ε\varepsilon-net and totally bounded metric space).

What is claimed about the cost, and what is not. Claimed: the proof below is carried out in ZF+DC\mathrm{ZF} + \mathrm{DC}, and DC\mathrm{DC} is used exactly once, at step 5.1. Not claimed: that DC\mathrm{DC} is necessary for the statement. Establishing necessity would mean separating the statement from ZF, which is an independence result, and this library proves none. The reason countable choice is not used instead is that the point added at each stage has to be at distance at least ε\varepsilon from the points already produced, so the set it is drawn from depends on the earlier stages; the first remark below spells that out.

Facts & Assumptions

Given: A sequentially compact metric space (X,d)(X,d), and the Axiom of Dependent Choice.

[L2]

(X,d)(X,d) is totally bounded when for every real ε>0\varepsilon > 0 there is a finite FXF \subseteq X, empty or listable, with X=yFB(y,ε)X = \bigcup_{y \in F} B(y,\varepsilon); equivalently, when for every real ε>0\varepsilon > 0 some finite list y0,,ym1y_0, \dots, y_{m-1} of points of XX satisfies: every xXx \in X has d(x,yi)<εd(x,y_i) < \varepsilon for some i<mi < m (Finite ε\varepsilon-net and totally bounded metric space, Open ball, closed ball and sphere in a metric space).

[L3]

Dependent choice: for a nonempty set SS, a relation RR on SS with every element RR-related to some element, and any aSa \in S, there is a sequence (tn)(t_n) in SS with t0=at_0 = a and tnRtn+1t_n \mathbin{R} t_{n+1} for every nn (The axiom of dependent choice: a relation in which every element is related to something admits an N\mathbb{N}-indexed chain).

[L4]

A convergent sequence is Cauchy: if ynjpy_{n_j} \to p then for every rational ε>0\varepsilon > 0 there is KK with d(ynj,ynl)<εd(y_{n_j}, y_{n_l}) < \varepsilon for all j,lKj,l \ge K (Every convergent sequence in a metric space is Cauchy, Cauchy sequence in a metric space).

[L6]

For every real η>0\eta > 0 there is a natural N1N \ge 1 with 1/N<η1/N < \eta, and 1/N1/N is a positive rational (For every ε>0\varepsilon > 0 in a complete ordered field there is a natural n1n \ge 1 with 1/n<ε1/n < \varepsilon, Every complete ordered field is Archimedean).

Proof

technique · contradiction
1.1

Suppose (X,d)(X,d) is sequentially compact and not totally bounded, and fix a real ε>0\varepsilon > 0 for which no finite subset of XX is an ε\varepsilon-net.

L1L2assume-contra
2.1

Then XX \ne \emptyset, since for X=X = \emptyset the empty set is an ε\varepsilon-net.

L2step 1.1
3.1

Let SS be the set of ε\varepsilon-separated finite tuples in XX, that is of functions t:mXt : m \to X with mNm \in \mathbb{N} and d(t(i),t(j))εd(t(i),t(j)) \ge \varepsilon whenever i<j<mi < j < m; the empty function, with m=0m = 0, lies in SS, so SS \ne \emptyset.

L5step 2.1
4.1

Let tRtt \mathbin{R} t' mean that t:m+1Xt' : m+1 \to X extends t:mXt : m \to X by one term with d(t(m),t(i))εd(t'(m), t(i)) \ge \varepsilon for every i<mi < m; then RR is a relation on SS and every tSt \in S is RR-related to some element of SS, because the finite set {t(0),,t(m1)}\{t(0), \dots, t(m-1)\} is not an ε\varepsilon-net, so some xXx \in X has d(x,t(i))εd(x,t(i)) \ge \varepsilon for every i<mi < m, and the extension of tt by xx lies in SS.

L2L5step 3.1
5.1

Dependent choice, applied to SS, to RR and to the empty function as starting point, yields a sequence (tn)(t_n) in SS with t0t_0 the empty function and tnRtn+1t_n \mathbin{R} t_{n+1} for every nn; this is the only appeal to a choice principle in the proof.

L3step 4.1
6.1

Each tnt_n has domain nn and tn+1t_{n+1} restricted to nn is tnt_n, both by induction on nn from the definition of RR; so yn:=tn+1(n)y_n := t_{n+1}(n) defines a sequence (yn)(y_n) in XX, and for i<ji < j both yi=tj+1(i)y_i = t_{j+1}(i) and yj=tj+1(j)y_j = t_{j+1}(j) hold, whence d(yi,yj)εd(y_i,y_j) \ge \varepsilon.

L5step 5.1
7.1

Sequential compactness gives a strictly increasing jnjj \mapsto n_j and pXp \in X with ynjpy_{n_j} \to p; that subsequence is therefore Cauchy, so, taking a natural N1N \ge 1 with 1/N<ε1/N < \varepsilon and testing the Cauchy condition at the positive rational 1/N1/N, there is KNK \in \mathbb{N} with d(ynj,ynl)<1/N<εd(y_{n_j}, y_{n_l}) < 1/N < \varepsilon for all j,lKj,l \ge K.

L1L4L6step 6.1
8.1

But nK<nK+1n_K < n_{K+1}, so step 6.1 gives d(ynK,ynK+1)εd(y_{n_K}, y_{n_{K+1}}) \ge \varepsilon, contradicting step 7.1; the assumption of step 1.1 is therefore untenable, every real ε>0\varepsilon > 0 admits a finite ε\varepsilon-net, and (X,d)(X,d) is totally bounded.

L2step 6.1step 7.1discharge-contradiction

Remarks

Why countable choice is not what this proof uses. A natural attempt is to apply ACω\mathrm{AC}_\omega to the family whose nn-th member is the set of ε\varepsilon-separated nn-tuples, each of which is nonempty by the argument of step 4.1. What that returns is one ε\varepsilon-separated nn-tuple for each nn, with no relation whatever between the tuple chosen at nn and the one chosen at n+1n+1: the tuples need not extend one another, need not share a single point, and nothing in the data assembles them into one ε\varepsilon-separated sequence. The relation RR of step 4.1 is precisely the coherence that is missing, and building a sequence along a relation is what The axiom of dependent choice: a relation in which every element is related to something admits an N\mathbb{N}-indexed chain is. This is an observation about the argument given here; it is not a proof that ACω\mathrm{AC}_\omega is insufficient for the theorem.

The passage to a rational ε\varepsilon in step 7.1. Convergence and the Cauchy condition are tested against rational ε\varepsilon in this library (Convergence of a sequence in a metric space: xkxx_k \to x iff d(xk,x)0d(x_k, x) \to 0 in R\mathbb{R}, Cauchy sequence in a metric space), while the ε\varepsilon of step 1.1 is an arbitrary positive real. The reciprocal form of the Archimedean property supplies a positive rational 1/N1/N below it (For every ε>0\varepsilon > 0 in a complete ordered field there is a natural n1n \ge 1 with 1/n<ε1/n < \varepsilon), and the contradiction is unaffected: a Cauchy estimate at 1/Nε1/N \le \varepsilon still contradicts a separation of at least ε\varepsilon.

This is the only implication on the page that costs dependent choice, and it is the reason For a metric space, compact, countably compact, limit point compact, sequentially compact, and complete together with totally bounded are all equivalent, given countable choice and dependent choice carries DC\mathrm{DC} among its hypotheses. The full accounting is What each implication between the compactness properties of a metric space costs: which are theorems of ZF, which use countable choice, and which use dependent choice.

Depends on

Used by

Dependency tree · next 3 levels

Direct dependencies and their dependencies through the next three levels: 90 results over 18 levels. An arrow runs from a result to what uses it, and this result sits at the bottom with a heavier outline. Click the chart to enlarge it.

Sources