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 complete, totally bounded metric space is compact, proved from countable choice used exactly once

Statement

Assume the Axiom of Countable Choice (The Axiom of Countable Choice (ACω\mathrm{AC}_\omega)). Let (X,d)(X,d) be a metric space (Metric space: d(x,y)=0d(x,y) = 0 iff x=yx = y, symmetry, and the triangle inequality; pseudometric and ultrametric) that is complete (Complete metric space: every Cauchy sequence converges in the space) and totally bounded (Finite ε\varepsilon-net and totally bounded metric space). Then (X,d)(X,d) is compact (Open cover, subcover, compact metric space, and compact subset of a metric space).

Where the axiom is spent, and why the weaker principle suffices. ACω\mathrm{AC}_\omega is used exactly once, at step 3.1, to fix one finite 1/(n+1)1/(n+1)-net together with a listing of it for every nNn \in \mathbb{N} at once. The family of sets being chosen from is written down before any selection is made and does not depend on the earlier selections, which is precisely the situation countable choice covers and dependent choice (The axiom of dependent choice: a relation in which every element is related to something admits an N\mathbb{N}-indexed chain) is not needed for. Everything after step 3.1 is canonical: at each stage the construction takes the least admissible index in the listing already fixed.

As always on this page, the claim is an upper bound on the cost of the proof given here, not an assertion that ACω\mathrm{AC}_\omega is necessary for the theorem.

Facts & Assumptions

Given: A complete, totally bounded metric space (X,d)(X,d), an open cover of it, and the Axiom of Countable Choice.

[L1]

(X,d)(X,d) is compact when every family U\mathcal{U} of open subsets of XX with union XX has a finite subfamily with union XX; UU open means every point of UU has a ball around it inside UU (Open cover, subcover, compact metric space, and compact subset of a metric space, 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, Open ball, closed ball and sphere in a metric space).

[L2]

(X,d)(X,d) is totally bounded: for every real δ>0\delta > 0 there is a finite FXF \subseteq X with X=yFB(y,δ)X = \bigcup_{y \in F} B(y,\delta), and a nonempty finite set can be listed (Finite ε\varepsilon-net and totally bounded metric space, Open cover, subcover, compact metric space, and compact subset of a metric space).

[L3]

Countable choice: for a family (En)nN(E_n)_{n \in \mathbb{N}} of nonempty sets there is a function nenn \mapsto e_n with enEne_n \in E_n for every nn (The Axiom of Countable Choice (ACω\mathrm{AC}_\omega)).

[L4]

Recursion: for a set AA, an element aAa \in A and a function f:AAf : A \to A there is a unique g:NAg : \mathbb{N} \to A with g(0)=ag(0) = a and g(n+1)=f(g(n))g(n+1) = f(g(n)); a stage-dependent rule is handled on A=N×ZA = \mathbb{N} \times Z, the first coordinate of g(n)g(n) then being nn (The recursion theorem, Finite sums and finite products, by recursion).

[L5]

Every nonempty subset of N\mathbb{N} has a least element, and the order of N\mathbb{N} is linear (The well-ordering principle, \le is a linear order on N\mathbb{N}).

[L6]

A function with domain a natural number all of whose values are nonempty sets has a choice function, in ZF (Every natural-number-indexed list of nonempty sets has a choice function on its family of values).

[L7]

(X,d)(X,d) is complete: every Cauchy sequence converges in XX; (cn)(c_n) is Cauchy when for every rational η>0\eta > 0 there is KK with d(cn,cm)<ηd(c_n,c_m) < \eta for n,mKn,m \ge K; and cnpc_n \to p when for every rational η>0\eta > 0 there is KK with d(cn,p)<ηd(c_n,p) < \eta for nKn \ge K (Complete metric space: every Cauchy sequence converges in the space, Cauchy sequence in a metric space, 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}, Sequences of reals: bounded, eventually, frequently, tails, subsequences).

[L8]

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; reciprocals of positives are positive and reverse the order (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, Inverses of positives are positive, and reciprocation reverses order).

Proof

technique · contradiction
1.1

Suppose (X,d)(X,d) is complete and totally bounded and that some family U\mathcal{U} of open subsets of XX with union XX has no finite subfamily with union XX; call a subset SXS \subseteq X finitely covered when some finite subfamily of U\mathcal{U} has union containing SS, so that XX itself is not finitely covered.

L1assume-contra
2.1

XX \ne \emptyset, since the empty subfamily of U\mathcal{U} has union \emptyset and would finitely cover an empty XX; consequently, for each nNn \in \mathbb{N}, the set EnE_n of pairs (m,f)(m,f) with mNm \in \mathbb{N} and f:σ(m)Xf : \sigma(m) \to X such that {f(0),,f(m)}\{f(0), \dots, f(m)\} is a finite 1/(n+1)1/(n+1)-net for XX is nonempty.

L2L8step 1.1
3.1

Countable choice applied to (En)nN(E_n)_{n \in \mathbb{N}} fixes, once and for all, a function n(mn,fn)n \mapsto (m_n, f_n) with (mn,fn)En(m_n,f_n) \in E_n; this is the single appeal to a choice principle in this proof, and the family (En)(E_n) was written down at step 2.1 before any of it was made.

L3step 2.1
4.1

If SXS \subseteq X is not finitely covered and nNn \in \mathbb{N}, then SB(fn(i),1/(n+1))S \cap B(f_n(i), 1/(n+1)) fails to be finitely covered for at least one imni \le m_n: otherwise finite choice applied to i{finite subfamilies of U covering SB(fn(i),1/(n+1))}i \mapsto \{\text{finite subfamilies of } \mathcal{U} \text{ covering } S \cap B(f_n(i), 1/(n+1))\} produces one such subfamily for each imni \le m_n, and the concatenation of those finitely many finite lists is a finite subfamily of U\mathcal{U} whose union contains S=imn(SB(fn(i),1/(n+1)))S = \bigcup_{i \le m_n} \big(S \cap B(f_n(i), 1/(n+1))\big).

L2L6step 3.1
5.1

Let i(n,S)i(n,S) be the least imni \le m_n with SB(fn(i),1/(n+1))S \cap B(f_n(i), 1/(n+1)) not finitely covered when such an ii exists, and 00 otherwise; recursion on N×P(X)\mathbb{N} \times \mathcal{P}(X) with starting value (0,X)(0, X) and rule F(n,S):=(n+1, SB(fn(i(n,S)),1/(n+1)))F(n,S) := \big(n+1,\ S \cap B(f_n(i(n,S)), 1/(n+1))\big) then produces g:NN×P(X)g : \mathbb{N} \to \mathbb{N} \times \mathcal{P}(X) whose first coordinate at nn is nn; write AnA_n for its second coordinate and cn:=fn(i(n,An))c_n := f_n(i(n,A_n)).

L4L5step 4.1
6.1

By induction, no AnA_n is finitely covered: A0=XA_0 = X is not, by step 1.1, and if AnA_n is not then step 4.1 supplies an admissible ii, so i(n,An)i(n,A_n) is one and An+1=AnB(cn,1/(n+1))A_{n+1} = A_n \cap B(c_n, 1/(n+1)) is not finitely covered either. In particular every AnA_n is nonempty, since the empty set is finitely covered by the empty subfamily; moreover An+1AnA_{n+1} \subseteq A_n and An+1B(cn,1/(n+1))A_{n+1} \subseteq B(c_n, 1/(n+1)).

step 4.1step 5.1
7.1

The sequence (cn)(c_n) is Cauchy: for mnm \ge n one has Am+1An+1B(cn,1/(n+1))A_{m+1} \subseteq A_{n+1} \subseteq B(c_n, 1/(n+1)) and Am+1B(cm,1/(m+1))A_{m+1} \subseteq B(c_m, 1/(m+1)), and taking zAm+1z \in A_{m+1} gives d(cn,cm)d(cn,z)+d(z,cm)<1/(n+1)+1/(m+1)2/(n+1)d(c_n,c_m) \le d(c_n,z) + d(z,c_m) < 1/(n+1) + 1/(m+1) \le 2/(n+1); so given a rational η>0\eta > 0, a natural N1N \ge 1 with 1/N<η/21/N < \eta/2 makes d(cn,cm)<ηd(c_n,c_m) < \eta for all n,mNn,m \ge N.

L8L9step 6.1
8.1

By completeness cnpc_n \to p for some pXp \in X; since U\mathcal{U} has union XX there is UUU \in \mathcal{U} with pUp \in U, and openness of UU gives a real r>0r > 0 with B(p,r)UB(p,r) \subseteq U.

L1L7step 7.1
9.1

Take a natural N1N \ge 1 with 1/N<r/21/N < r/2, then KNK \in \mathbb{N} with d(cn,p)<1/Nd(c_n,p) < 1/N for all nKn \ge K, and let nn be whichever of NN and KK is the greater; then 1/(n+1)<1/N<r/21/(n+1) < 1/N < r/2 and d(cn,p)<r/2d(c_n,p) < r/2, so every yAn+1B(cn,1/(n+1))y \in A_{n+1} \subseteq B(c_n,1/(n+1)) satisfies d(y,p)d(y,cn)+d(cn,p)<rd(y,p) \le d(y,c_n) + d(c_n,p) < r, that is An+1B(p,r)UA_{n+1} \subseteq B(p,r) \subseteq U.

L5L7L8L9step 8.1
10.1

So the one-member subfamily {U}\{U\} of U\mathcal{U} has union containing An+1A_{n+1}, making An+1A_{n+1} finitely covered and contradicting step 6.1; the assumption of step 1.1 therefore fails, every family of open sets with union XX has a finite subfamily with union XX, and (X,d)(X,d) is compact.

L1step 6.1step 9.1discharge-contradiction

Remarks

Why the nets have to be chosen with their listings. Total boundedness asserts that a finite 1/(n+1)1/(n+1)-net exists for each nn; it names none, and a bare net is a set, which carries no order in which its points may be scanned. The construction needs both: a net for each nn, so that the sets AnA_n shrink, and a listing of it, so that "the least admissible index" is meaningful. That is why the chosen object at step 3.1 is the pair (mn,fn)(m_n, f_n) and not the net alone.

Non-dependent, and that is the whole point. The sets EnE_n of step 2.1 depend on nn and on (X,d)(X,d), and on nothing that the construction produces. Had the net at stage n+1n+1 been required to depend on AnA_n — for instance a net of the set AnA_n rather than of XX — the selection would have been dependent and countable choice would not have licensed it; the cost would then have been the dependent choice of The axiom of dependent choice: a relation in which every element is related to something admits an N\mathbb{N}-indexed chain, as in A sequentially compact metric space is totally bounded, proved from the axiom of dependent choice. Keeping the nets fixed in advance and intersecting with balls of XX is what holds the price down.

Both hypotheses are needed. A totally bounded space that is not complete need not be compact (FALSE: a totally bounded metric space is compact, The open interval (0,1)(0,1) is totally bounded and not compact, the cover by the intervals (1/(k+2),1)(1/(k+2), 1) having no finite subcover ), and a complete space that is not totally bounded need not be compact either, R\mathbb{R} with its usual metric being complete and having no finite 11-net.

Depends on

Used by

Dependency tree · next 3 levels

Direct dependencies and their dependencies through the next three levels: 96 results over 20 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