Alphabeta Math
TheoremStatement: AI-adaptedProof: AI-generatedSession-authored (Fable 5 assisted)precheck passjudge pass (z-ai/glm-5.2)audited 2026-07-29
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.

Assuming countable choice, every perfectly normal space is completely normal: separated sets in a normal space whose open sets are all FσF_\sigma can be separated by disjoint open sets

Statement

Assume the Axiom of Countable Choice (The Axiom of Countable Choice (ACω\mathrm{AC}_\omega)). Let (X,T)(X, \mathcal{T}) be a perfectly normal space (Completely normal (T5T_5) and perfectly normal (T6T_6) spaces): XX is normal (Normal spaces and T4T_4 spaces, with the source disagreement over whether normality includes T1T_1 stated explicitly) and every closed subset of XX is a GδG_\delta, equivalently every open subset of XX is an FσF_\sigma (GδG_\delta and FσF_\sigma subsets of a topological space, agreeing with the real-line notion). Then XX is completely normal: any two separated sets A,BXA, B \subseteq X (Separated sets: AB=AB=\overline{A} \cap B = A \cap \overline{B} = \varnothing) admit disjoint open UAU \supseteq A and VBV \supseteq B.

Consequently T6T_6 implies T5T_5.

No continuous function is constructed anywhere in the proof, and in particular Urysohn's lemma is not used. All that is consumed is normality, applied once to each member of a countable family of closed sets, and the FσF_\sigma presentation of two open sets.

Where the choice principle is spent, and why it is not removable as written. Step 4.1 selects, for each nNn \in \mathbb{N} at once, one open set UnU_n out of the nonempty family that normality provides for the closed set FnF_n, and likewise one VnV_n; normality is an existence statement and supplies no rule for singling out a member, so extracting the two sequences is an application of ACω\mathrm{AC}_\omega and of nothing stronger. The hypothesis is stated in the theorem rather than hidden in the proof, as this library does everywhere.

Facts & Assumptions

Given: A perfectly normal space (X,T)(X,\mathcal{T}) and separated sets A,BXA, B \subseteq X, so that AB=AB=\overline{A} \cap B = A \cap \overline{B} = \varnothing.

[A1]

AA and BB are separated: AB=\overline{A} \cap B = \varnothing and AB=A \cap \overline{B} = \varnothing (Separated sets: AB=AB=\overline{A} \cap B = A \cap \overline{B} = \varnothing).

[A2]

Every open subset of XX is an FσF_\sigma: it is nNCn\bigcup_{n \in \mathbb{N}} C_n for some sequence of closed sets CnC_n (Completely normal (T5T_5) and perfectly normal (T6T_6) spaces, GδG_\delta and FσF_\sigma subsets of a topological space, agreeing with the real-line notion, Finite, countably infinite, countable, uncountable).

[A3]

ACω\mathrm{AC}_\omega: for a family of nonempty sets indexed by N\mathbb{N} there is a function choosing a member of each (The Axiom of Countable Choice (ACω\mathrm{AC}_\omega)).

[L3]

A union of finitely many closed sets is closed by iterating (C3), an arbitrary union of open sets is open by (T2), and an intersection of two open sets is open by (T3) (Topology on a set, open and closed sets, clopen sets, the closed-set axiomatisation, and the coarser/finer comparison).

[L4]

For all n,mNn, m \in \mathbb{N} exactly one of n<mn < m, n=mn = m, m<nm < n holds (Trichotomy of the order on N\mathbb{N}).

Proof

technique · direct
1.1

AXBA \subseteq X \setminus \overline{B} and BXAB \subseteq X \setminus \overline{A}, and both of these sets are open.

A1L2
2.1

By [A2] fix sequences of closed sets with XB=nNFnX \setminus \overline{B} = \bigcup_{n \in \mathbb{N}} F_n and XA=nNGnX \setminus \overline{A} = \bigcup_{n \in \mathbb{N}} G_n.

step 1.1A2choose
3.1

For every nn the closed sets FnF_n and B\overline{B} are disjoint, since FnXBF_n \subseteq X \setminus \overline{B}; likewise GnG_n and A\overline{A} are disjoint closed sets.

step 2.1L2
4.1

By [L1] the set of open WFnW \supseteq F_n with WB=\overline{W} \cap \overline{B} = \varnothing is nonempty for each nn, and likewise the set of open WGnW' \supseteq G_n with WA=\overline{W'} \cap \overline{A} = \varnothing; so [A3] supplies sequences (Un)nN(U_n)_{n \in \mathbb{N}} and (Vn)nN(V_n)_{n \in \mathbb{N}} of open sets with FnUnF_n \subseteq U_n, UnB=\overline{U_n} \cap \overline{B} = \varnothing, GnVnG_n \subseteq V_n and VnA=\overline{V_n} \cap \overline{A} = \varnothing for every nn.

step 3.1A3L1choose
5.1

Define U:=nN(UninVi)U := \bigcup_{n \in \mathbb{N}} \big(U_n \setminus \bigcup_{i \le n} \overline{V_i}\big) and V:=nN(VnjnUj)V := \bigcup_{n \in \mathbb{N}} \big(V_n \setminus \bigcup_{j \le n} \overline{U_j}\big).

step 4.1construct
6.1

UU and VV are open: for each nn the set inVi\bigcup_{i \le n} \overline{V_i} is a union of finitely many closed sets, hence closed, so its complement is open and UninViU_n \setminus \bigcup_{i \le n} \overline{V_i} is an intersection of two open sets; the union over nn is then open.

step 5.1L2L3
6.2

AUA \subseteq U: given aAa \in A, step 1.1 and step 2.1 put aa in some FnUnF_n \subseteq U_n, while aAa \in \overline{A} and ViA=\overline{V_i} \cap \overline{A} = \varnothing give aVia \notin \overline{V_i} for every ii; hence aUninViUa \in U_n \setminus \bigcup_{i \le n} \overline{V_i} \subseteq U.

step 1.1step 2.1step 4.1step 5.1L2
6.3

BVB \subseteq V: given bBb \in B, step 1.1 and step 2.1 put bb in some GmVmG_m \subseteq V_m, while bBb \in \overline{B} and UjB=\overline{U_j} \cap \overline{B} = \varnothing give bUjb \notin \overline{U_j} for every jj; hence bVmjmUjVb \in V_m \setminus \bigcup_{j \le m} \overline{U_j} \subseteq V.

step 1.1step 2.1step 4.1step 5.1L2
6.4

Suppose xUVx \in U \cap V; then by step 5.1 there are n,mNn, m \in \mathbb{N} with xUnx \in U_n, xVix \notin \overline{V_i} for all ini \le n, xVmx \in V_m, and xUjx \notin \overline{U_j} for all jmj \le m.

step 5.1assume-hyp
7.1

If nmn \le m in step 6.4 then j:=nj := n satisfies jmj \le m, so xUnx \notin \overline{U_n}; but xUnUnx \in U_n \subseteq \overline{U_n}, which is impossible.

step 6.4L2
7.2

If m<nm < n in step 6.4 then i:=mi := m satisfies ini \le n, so xVmx \notin \overline{V_m}; but xVmVmx \in V_m \subseteq \overline{V_m}, which is impossible.

step 6.4L2
8.1

By [L4] one of nmn \le m and m<nm < n holds, so steps 7.1 and 7.2 exclude every case and no such xx exists: UV=U \cap V = \varnothing.

step 7.1step 7.2L4
9.1

By steps 6.1, 6.2, 6.3 and 8.1 the sets UU and VV are disjoint open sets containing AA and BB respectively; since AA and BB were an arbitrary separated pair, XX is completely normal, and with the hypothesis T1T_1 this reads T6T_6 implies T5T_5.

step 6.1step 6.2step 6.3step 8.1

Remarks

  • The subtraction of the earlier closures is the entire trick. Each UninViU_n \setminus \bigcup_{i \le n} \overline{V_i} is still large enough to catch the part of AA that FnF_n covers, because no point of AA lies in any Vi\overline{V_i}; and it is small enough that the two unions cannot meet, because a putative common point would be inside a UnU_n that a later stage of VV has already removed, or inside a VmV_m that a later stage of UU has removed. The comparison nmn \le m or m<nm < n is what decides which of the two it is.

  • Only the two closures A\overline{A} and B\overline{B} are used, never the sets AA and BB themselves beyond membership, which is why the hypothesis is exactly separation and not disjointness. For disjoint sets that are not separated the argument breaks at step 6.2.

  • The converse is not proved here and is not asserted. Perfect normality asks a countability condition of every closed set that complete normality never mentions, so the two are not the same hypothesis; but no witness separating them is exhibited in this library, and nothing above claims one exists.

  • The hereditary reading is not used. Complete normality is equivalent to the normality of every subspace, and some texts prove this theorem in that language; the argument above works directly with the separated-sets definition and never passes to a subspace.

Depends on

Used by

Dependency tree · next 3 levels

Direct dependencies and their dependencies through the next three levels: 98 results over 26 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