Alphabeta Math
TheoremStatement: Literature-sourcedProof: AI-adaptedPipeline-generatedprecheck passaudited 2026-09-14
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.

Numerable principal bundles are classified by maps to BG

Statement

Assume AC. For a well-pointed topological group G of CW type and a CGWH base X, pullback of Milnor's bundle induces a bijection

[X,BG]  BunGnum(X),[f][fEG],

where the right side consists of isomorphism classes of numerable right principal G-bundles. A locally trivial bundle over a paracompact base is covered only after a separate theorem supplies numerability.

Facts & Assumptions

[F1]

Milnor's EGBG is a numerable principal bundle; EG is contractible; the even and odd coordinate embeddings are equivariantly homotopic to its identity; and disjoint-support interpolation after those embeddings is a continuous equivariant homotopy (Milnor's join model is a contractible free G-space).

[F2]

Pullback preserves principal-bundle charts (Associated bundle is locally trivial and functorial under pullback), and the pullback of a support-subordinate partition is support-subordinate by inverse-image functoriality of support.

[F3]

The lifting construction for a numerable bundle is made from chart transports and therefore commutes with a right principal action (Numerable fiber bundles are hurewicz fibrations).

[F4]

A continuous equivariant map between principal G-bundles over the same base is a bundle isomorphism, as follows in principal charts from the torsor condition (Principal g bundle and associated fiber bundle).

[A1]

AC chooses members and chart data for arbitrary indexed families in the countabilization below. Finite supplied numerations do not require this use (The Axiom of Choice).

Proof

Given: X,G and [A1] as in the statement.

1.1

First let π:PX be numerable with an arbitrary indexed numeration (ρi)iI subordinate to principal charts Ui. For each nonempty finite SI, define

A1

uS(x)=max(0,miniSρi(x)supjSρj(x)),

where the supremum of an empty family is 0. Near each point only finitely many ρi can be nonzero, so the displayed supremum is locally a finite maximum and uS is continuous. At a point, let S be the finite set of indices attaining the largest positive value; then uS>0. If SS have the same cardinality, their cozero sets are disjoint, since indices in SS and SS would otherwise have to be strictly larger than one another. [A1]

1.2

The assignment depends only on the homotopy class of f. If H:X×IBG joins f0 to f1, then HEG is numerable by [F1, F2]. Apply the lifting function of [F3] to the paths tH(x,t). Holding the principal group coordinate in every chart makes endpoint transport an equivariant map from the restriction over X×{0} to that over X×{1}. It is a bundle isomorphism by [F4]. Thus f0EGf1EG.

F1F2F3F4
1.3

Conversely, suppose f0EGf1EG and identify both with one principal bundle P. Projection to the EG coordinate gives equivariant maps q0,q1:PEG covering f0,f1. By [F1], equivariantly deform q0 to a map q0ev supported in even coordinates and q1 to q1odd supported in odd coordinates. Their disjoint supports make

F1

K(p,t)=(1t)q0ev(p)+tq1odd(p)

a well-defined continuous equivariant map P×IEG. Passing to orbits gives a homotopy between the two deformed base maps. Concatenating with the orbit homotopies furnished by [F1] proves f0f1. [F1]

2.1

For m1, set wm=S=muS. These sums are locally finite, their cozero sets Vm cover X, and Vm is the disjoint union of the cozero sets of the uS with S=m. Each such piece lies in every Ui with iS. Use [A1] to choose one i(S)S; restricting its section and patching over the disjoint pieces gives a section sm:VmP. Normalize the wm, then apply the threshold construction of the Milnor theorem with positive numbers summing to less than one. We obtain a countable locally finite partition (λm)m1 with suppλmVm.

A1F1step 1.1
3.1

For pP over x, whenever λm(x)>0 write uniquely p=sm(x)am(p). Define

F1step 2.1

Φ(p)=m1λm(x)am(p)EG.

Only finitely many terms occur locally. Support containment makes the quotient formula continuous even where a label ceases to be defined, and on such a neighborhood it factors through one finite-join quotient. Thus Φ is continuous. It is equivariant because am(pg)=am(p)g, and it descends to a map f:XBG. [F1, step 2.1]

4.1

The map p(π(p),Φ(p)) is an equivariant map PfEG over X. On each fiber it is a map of right G-torsors and is therefore bijective. In principal charts it has the form (x,g)(x,c(x)g), whose inverse is (x,h)(x,c(x)1h); hence [F4] makes it a bundle isomorphism. Every numerable bundle is therefore pulled back from Milnor's bundle.

F4step 3.1
5.1

Steps 1.2, 1.3, and 4.1 prove well-definedness, injectivity, and surjectivity of the displayed map. If X=, both sides are singletons. If the given numeration is finite, its countabilization and all chart choices in Steps 1.1--3.1 are finite; for arbitrary index sets, [A1] is exactly the declared choice use. The theorem makes no claim that paracompactness implies numerability.

A1step 1.2step 1.3step 4.1

Depends on

Used by

Dependency tree · two levels

21 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