Alphabeta Math
Pipeline-generated
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.

✓ 13 results · all verified · 7 also independently AI-judged
Every result on this page is machine-checked by a proof checker and read in full and owner-audited; the judge is an additional, independent cross-model AI review of the proofs. The 6 not AI-judged were verified by owner audit (typically over a confirmed judge false positive), not failures.

Pontryagin Thom and Framed Cobordism

1 · Prerequisites

2 · Summary

This page develops the Pontryagin–Thom construction and the framed cobordism relation on closed submanifolds of a closed smooth manifold X. A framing of a codimension-k submanifold is an actual trivialization of its quotient normal bundle, never merely a stable one, and a framed cobordism is built from a compact neat submanifold of X×I with product ends carrying framings that restrict to the two given framings; the collars and product ends are part of the data, so reflexivity, symmetry and gluing are literal and no boundary sign is left implicit. Countable choice ACω is inherited from the smooth normal-bundle and compatible-chart machinery and is tracked through the theory; the Hopf-bundle examples explicitly assume full AC for the supplied numerable-bundle lifting theorem.

The forward construction collapses a tubular neighbourhood of a framed submanifold to the Thom space of its normal bundle; a framing identifies that Thom target with N+∧Sk, and the composite with the based projection is the Pontryagin–Thom map X→Sk. Its homotopy class is independent of the tube, the metric and the radius, and a framed cobordism supplies an explicit homotopy between the collapse maps of its ends. In the reverse direction the regular preimage of a map to Sk at a regular value is framed by the differential. For k≥1, its class is independent of the regular value and the positive basis; homotopic maps with a common regular value give framed-cobordant preimages. The two constructions are checked against each other on the nose: the regular preimage of the normalized smooth Pontryagin–Thom representative at its centre recovers the original framed submanifold with its framing, and the Pontryagin–Thom map of a regular preimage is homotopic to the original map.

These inverse checks assemble into the fixed-codimension theorem: for n≥k≥1 the collapse and the regular-preimage constructions define mutually inverse bijections between framed cobordism classes of closed framed (n−k)-submanifolds of Sn and πn(Sk), equivalently free homotopy classes of maps Sn→Sk. Passing to codimension-independent statements, equatorial stabilization of framed submanifolds defines a directed system whose colimit is the framed bordism group Ωdfr, and stabilization suspends the Pontryagin–Thom map; the levelwise bijections therefore induce the stable Pontryagin–Thom isomorphism Ωdfr→πds, which is additive and hence an isomorphism of abelian groups for disjoint union on the left and addition on the right. A closing remark keeps three structures apart that are easy to conflate: actual normal framings, stable normal framings and stable tangential framings.

3 · Logical flowchart

4 · Definitions, theorems and proofs

DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6.1-sol)Open item page →

Framings of a normal bundle

Definition

Assume ACω (The Axiom of Countable Choice (ACω)). Let S⊆X be a closed embedded smooth submanifold of a smooth manifold X without boundary, with normal bundle ν(S)=TX∣S/TS of rank k, a smooth vector bundle over S by Assuming countable choice, normal and conormal bundles are smooth vector bundles (Normal and conormal bundles of an embedded submanifold, Smooth embeddings).

A framing of S in X is a smooth bundle isomorphism φ:ν(S)⟶S×Rk over idS (Smooth vector bundles, rank, fibres, and trivial bundles); equivalently a trivialization of ν(S), equivalently a global frame of its sections, and in Milnor's metric language the same thing as a framing of the orthogonal complement TS⊥. A framed submanifold is a pair (S,φ). When a Riemannian metric on X is supplied, the orthogonal-complement model of the normal bundle is canonically identified with ν(S) by Assuming countable choice, an ambient metric identifies the two normal bundles, which is how Milnor's framings are read in the metric-free quotient convention used here; a change of ambient metric changes that comparison but not the framing data itself.

Rank k=0 and S=∅ are included (Open cover, subcover, and compact topological space; a compact subset is a subspace that is compact in its own right): a framing of a rank-zero bundle is the unique bundle isomorphism onto S×R0=S, and the empty framing is unique. A framing is an actual trivialization of ν(S), never merely a stable isomorphism of it: adding trivial summands to a normal bundle is a different construction, and the stable and unstable notions are kept apart throughout (Smooth vector bundles, rank, fibres, and trivial bundles).

The countable-choice hypothesis is inherited solely from the smooth normal-bundle structure of Assuming countable choice, normal and conormal bundles are smooth vector bundles; this definition itself selects nothing and proves nothing. The symbol ν(S) always denotes the quotient normal bundle TX∣S/TS, and the rank-k trivialization that fixes the identification of the normal fibres with Rk is part of the data, not a choice made afterwards.

DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6.1-sol)Open item page →

Framed cobordism of framed submanifolds

Definition

Assume ACω (The Axiom of Countable Choice (ACω)). Let X be a closed smooth manifold and k≥0 (Smooth manifolds and their smooth charts, Open cover, subcover, and compact topological space; a compact subset is a subspace that is compact in its own right). A framed cobordism from a closed framed codimension-k submanifold (N0,φ0) of X to another (N1,φ1) is data (W,ε,Ψ) consisting of

  • a compact neat embedded submanifold W⊆X×I of dimension dim⁡X−k+1 with ∂W=N0×{0}⊔N1×{1}, the Ni read in the slice X×{i} (Neat submanifolds of a manifold with boundary, Smooth embeddings; here I=[0,1] with its product smooth structure and X has been identified with X×{i});
  • a number ε∈(0,12) such that the ends of W are exactly the products W∩(X×[0,ε))=N0×[0,ε),W∩(X×(1−ε,1])=N1×(1−ε,1]; equivalently, the product collar embeddings θ0:N0×[0,ε)→W, θ0(x,s)=(x,s) and θ1:N1×(1−ε,1]→W, θ1(x,s)=(x,s) are part of the data, with θi(x,i)=(x,i) and images exactly the two ends;
  • a framing Ψ:ν(W⊆X×I)→W×Rk (Framings of a normal bundle) which, over each end collar Ni×Θi (where Θ0=[0,ε) and Θ1=(1−ε,1]), is the pullback of φi along the product projection, under the canonical identification ν(W⊆X×I)∣Ni×Θi≅pr⁡Ni∗ν(Ni⊆X) induced by the product structure: along the whole collar the I-direction is tangent to W, so the quotient normal of W in X×I restricts there to the quotient normal of Ni in X. In particular, at the end slice t=i the restriction of Ψ corresponds to φi; requiring the constancy over the whole collar is Milnor's normalisation ui(x,t)=(vi(x),0).

Two closed framed codimension-k submanifolds of X are framed cobordant when such data exist. The relation is introduced here only as a relation; that it is reflexive, symmetric and transitive is proved in Framed cobordism is an equivalence relation.

No orientation of X or of the Ni is used, and no direction of the normal bundle is singled out: all signs are carried by the actual framings φ0,φ1,Ψ. The product ends and the constant framings on them are data, not choices made afterwards, so the restriction of Ψ to each end is a literal equality with φi, with no implicit inward-normal sign and no implicit straightening of a general collar; this is Milnor's definition of cobordism within M, in which the subset N0×[0,ε)∪N1×(1−ε,1] extends to W. The empty manifold is allowed as N0, as N1 and as W, and k=0 is allowed; for k=0 the normal bundles are rank zero, the framings are unique, and the condition on Ψ is vacuous. A framed cobordism (W,ε,Ψ) also yields an (unoriented) bordism (W,θ0,θ1) in the sense of Unoriented smooth cobordism of closed manifolds after rescaling θi to the standard widths, since W is a compact smooth manifold with boundary and the θi are collars onto the two boundary parts. The countable-choice hypothesis is inherited from the smooth normal-bundle structure through Framings of a normal bundle; the definition itself selects nothing.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)Open item page →

Framed cobordism is an equivalence relation

Statement

Assume ACω (The Axiom of Countable Choice (ACω)). For every closed smooth manifold X and every k≥0, framed cobordism of closed framed codimension-k submanifolds of X (Framed cobordism of framed submanifolds) is an equivalence relation.

Reflexivity is realised by the cylinder N×I⊆X×I together with the pullback of φ along pr⁡N; symmetry is realised by the reflection ρ(x,t)=(x,1−t) of X×I with the framing transported across dρ; transitivity is realised by gluing two framed cobordisms along their common product end, the framings agreeing on the whole overlap collar.

Facts & Assumptions

Given: A closed smooth manifold X, an integer k≥0, and the definition of a framed cobordism (W,ε,Ψ) from (N0,φ0) to (N1,φ1).

[F1]

A framed cobordism consists of a compact neat embedded submanifold W⊆X×I with ∂W=N0×{0}⊔N1×{1}, product ends W∩(X×[0,ε))=N0×[0,ε) and W∩(X×(1−ε,1])=N1×(1−ε,1], and a framing Ψ:ν(W⊆X×I)→W×Rk that on each end collar is the pullback of φi under the canonical identification ν(W)∣Ni×Θi≅pr⁡∗ν(Ni⊆X) (Framed cobordism of framed submanifolds, Framings of a normal bundle, Neat submanifolds of a manifold with boundary).

[F2]

If N⊆X is a closed embedded submanifold then N×I⊆X×I is a closed embedded submanifold for the product smooth structures, and the product structure identifies ν(N×I⊆X×I) with pr⁡N∗ν(N⊆X) because the I-direction is tangent to N×I (Products of smooth manifolds have a canonical product smooth structure, Smooth manifolds and their smooth charts).

[F3]

The maps ρ(x,t)=(x,1−t) of X×I to itself, σ1(x,t)=(x,t/2) of X×I onto X×[0,12] and σ2(x,t)=(x,(t+1)/2) of X×I onto X×[12,1] are diffeomorphisms, and a diffeomorphism carries neat embedded submanifolds onto neat embedded submanifolds; its differential, by the chain rule, intertwines tangent and normal quotients and preserves the product splittings T(X×I)=TX⊕TI (Diffeomorphisms and local diffeomorphisms of manifolds, The chain rule for differentials of smooth maps).

[F5]

An equivalence relation on a set is a reflexive, symmetric and transitive relation (Equivalence relation, equivalence class, and the quotient set A/∼).

Proof

1.1F1F2F4construct

(Reflexivity.) Let (N,φ) be a closed framed codimension-k submanifold of X. Put W:=N×I⊆X×I and ε:=14. By [F2] and [F4], W is a compact embedded submanifold; its boundary in X×I is N×{0}⊔N×{1}=∂W, and W is transverse to ∂(X×I)=X×{0,1} because TW contains the I-direction, so W is neat. Its ends are the literal products N×[0,14) and N×(34,1]. By [F2] the normal bundle ν(W⊆X×I) is canonically pr⁡N∗ν(N⊆X); let Ψ be the pullback of φ along pr⁡N under this identification. Then Ψ is a smooth bundle isomorphism and over each end collar it is the pullback of φ, so (W,14,Ψ) is a framed cobordism from (N,φ) to (N,φ).

1.2F1F3construct

(Symmetry.) Let (W,ε,Ψ) be a framed cobordism from (N0,φ0) to (N1,φ1). Put ρ(x,t):=(x,1−t) and W′:=ρ(W). By [F3] W′ is a compact neat embedded submanifold of X×I with ∂W′=N1×{0}⊔N0×{1}, and its ends are the literal products ρ(N1×(1−ε,1])=N1×[0,ε) and ρ(N0×[0,ε))=N0×(1−ε,1]. The differential dρ is the identity on the TX factor and multiplication by −1 on the TI factor; along an end collar the I-direction is tangent to W and to W′, so dρ induces, over ρ, the identity map ν(Ni⊆X)→ν(Ni⊆X) of the canonical product identifications. Define Ψ′ over W′ by Ψρ(p)′:=Ψp∘(dρp‾)−1, where dρp‾:ν(W)p→ν(W′)ρ(p) is the quotient isomorphism induced by dρp. Then Ψ′ is a smooth bundle isomorphism ν(W′⊆X×I)→W′×Rk which on the collar N1×[0,ε) is the pullback of φ1 and on N0×(1−ε,1] the pullback of φ0. Hence (W′,ε,Ψ′) is a framed cobordism from (N1,φ1) to (N0,φ0).

1.3F1F3construct

(Transitivity, construction.) Let (W1,ε1,Ψ1) be a framed cobordism from (N0,φ0) to (N1,φ1) and (W2,ε2,Ψ2) one from (N1,φ1) to (N2,φ2). Put ε:=min⁡(ε1,ε2), define σ1(x,t):=(x,t/2) and σ2(x,t):=(x,(t+1)/2), and set W1∗:=σ1(W1)⊆X×[0,12], W2∗:=σ2(W2)⊆X×[12,1] and W:=W1∗∪W2∗. By [F3] each Wi∗ is a compact neat embedded submanifold, with ∂W1∗=N0×{0}⊔N1×{12} and ∂W2∗=N1×{12}⊔N2×{1}.

2.1F1F3step 1.2step 1.3

(Transitivity: W is a neat submanifold carrying a framing.) Near t=12 one has W1∗∩(X×(12−ε2,12])=N1×(12−ε2,12] and W2∗∩(X×[12,12+ε2))=N1×[12,12+ε2), so W∩(X×(12−ε2,12+ε2))=N1×(12−ε2,12+ε2) is a product piece. Every point of W with t≠12 lies in W1∗∩(X×[0,12)) or in W2∗∩(X×(12,1]), which are open pieces of the smooth submanifolds Wi∗; those two pieces and the product piece cover W, and on their overlaps (subsets of the product piece) the three descriptions agree. Hence W is a compact neat embedded submanifold of X×I with ∂W=N0×{0}⊔N2×{1} and with the literal product ends N0×[0,ε2) and N2×(1−ε2,1]. [F1, F3, step 1.3] Transport Ψ1 across σ1 and Ψ2 across σ2 as in step 1.2. Since σi preserves the splitting T(X×I)=TX⊕TI up to a positive scale in TI, the transported framings are smooth bundle isomorphisms ν(Wi∗⊆X×I)→Wi∗×Rk, the first equal to the pullback of φ1 on the collar N1×(12−ε2,12] and the second equal to the pullback of φ1 on N1×[12,12+ε2). These two collars cover the overlap region just described, where both transported framings are the pullback of φ1 under the canonical identification ν(W)∣N1×Θ≅pr⁡∗ν(N1⊆X); away from the overlap each is smooth. Hence they define one smooth bundle isomorphism Ψ:ν(W⊆X×I)→W×Rk, which on the outer ends is the pullback of φ0 and of φ2. Therefore (W,ε2,Ψ) is a framed cobordism from (N0,φ0) to (N2,φ2).

3.1F1F5step 1.1step 1.2step 2.1∎

(Conclusion.) Steps 1.1, 1.2 and 2.1 exhibit reflexivity, symmetry and transitivity of framed cobordism for arbitrary closed X, k≥0 and framed submanifolds, including the empty manifold and the rank-zero case k=0, where all framings are unique. By [F5] framed cobordism is an equivalence relation. No orientation, no metric and no choice beyond the inherited ACω is used: all constructions are explicit, and the only choice-dependent input is the smooth normal-bundle structure of [F1].

PropositionStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)Open item page →

A framing identifies the Thom target with a sphere smash product

Statement

Assume ACω (The Axiom of Countable Choice (ACω) is inherited through the normal-bundle structure). Let X be a closed smooth manifold and let (N,φ) be a closed framed codimension-k submanifold of X, k≥0, with normal bundle ν=ν(N⊆X) (Framings of a normal bundle).

Then φ induces a based homeomorphism Φφ:Th⁡(ν)⟶N+∧Sk, natural in the framed data, and the composite of any based map c:X+→Th⁡(ν) with the based projection N+∧Sk→S0∧Sk=Sk is a based map X+→Sk. The homeomorphism is independent of the metric used to form Th⁡(ν) up to the canonical radial homeomorphisms of Metric independence of the Thom space. For k=0, Th⁡(ν)=N+≅N+∧S0 and the sphere-valued target is S0. For N=∅, the Thom space and N+∧Sk are one-point spaces, and the composite X+→Sk is constant at the basepoint.

Facts & Assumptions

Given: A closed framed codimension-k submanifold (N,φ) of the closed smooth manifold X, its quotient normal bundle ν=TX∣N/TN, and a metric h on ν when a Thom space is formed.

[F1]

A framing is a smooth bundle isomorphism φ:ν→N×Rk over idN; rank zero and N=∅ are included and the framing is then unique (Framings of a normal bundle).

[F2]

With the product metric and supplied trivialization, Th⁡(B×Rr)≅B+∧Sr naturally in B, including the rank-zero and empty-base cases (Trivial Thom spaces as suspension smash products).

[F3]

The Thom space Th⁡h(E)=Dh(E)/Sh(E) is formed from the metric disk and sphere bundles, with X/∅=X+ convention and the nonbasepoint stratum the open disk bundle (Disk bundle, sphere bundle, and Thom space: the differential topology interface).

[F4]

Different metrics on E are compared by a canonical based homeomorphism, and these comparisons compose exactly (Metric independence of the Thom space).

[F5]

For based CGWH spaces the smash product is the kified quotient of the product by the wedge, it is associative, symmetric and unital up to canonical based homeomorphisms S0∧Z≅Z, and these homeomorphisms satisfy the usual coherence identities (Smash product of based spaces, Canonical associativity, symmetry, and unit maps for smash products, Compactly generated based spaces and well-pointed objects).

Proof

1.1F1F3construct

(The framing carries one Thom model to the other.) Fix a metric h on ν and let φ∗h be the metric on N×Rk obtained by transporting h across the bundle isomorphism φ. Since φ is a fibrewise linear homeomorphism over idN, it maps Dh(ν) onto Dφ∗h(N×Rk) and Sh(ν) onto Sφ∗h(N×Rk); passing to the quotients it induces a homeomorphism Th⁡h(ν)→Th⁡φ∗h(N×Rk) carrying basepoint to basepoint. This map is based and functorial: for N=∅ it is the unique map of one-point spaces, while for k=0 it is the identity on N+.

2.1F2F4step 1.1

(The trivial model and metric independence.) By [F2] applied to the trivial bundle N×Rk there is a canonical based homeomorphism Th⁡prod(N×Rk)≅N+∧Sk for the product metric, natural in N. Composing with step 1.1 and the exact metric comparison r:Th⁡φ∗h≅Th⁡prod from [F4] gives a based homeomorphism Φφ:Th⁡h(ν)→N+∧Sk. If h,h′ are two metrics on ν, the two composites differ by the canonical radial homeomorphism r′∘r−1 of [F4], which is exactly the asserted independence: the construction is natural in the framing, because a bundle isomorphism intertwines the transported metrics and hence the two routes through the framing isomorphism and metric comparison. For k=0, N×R0=N and [F2] gives Th⁡(N)=N+≅N+∧S0; for N=∅, all four spaces are the one-point based space and all maps are the identity.

3.1F5step 1.1step 2.1

(The sphere-valued collapse.) Let p:N+∧Sk→S0∧Sk be the smash of the based collapse N+→S0 (which sends N to the nonbasepoint) with idSk, followed by the canonical unitality homeomorphism S0∧Sk≅Sk of [F5]; the composite p is a based map. For any based c:X+→Th⁡(ν), the composite p∘Φφ∘c:X+→Sk is based because each factor is based, and it is independent of which unitality homeomorphism is used by the coherence clause of [F5].

4.1F1F2F3F4F5step 1.1step 2.1step 3.1∎

(Conclusion.) Steps 1.1-3.1 construct the based homeomorphism Φφ, prove its naturality in the framed data, its metric independence up to the canonical radial homeomorphism, and the based sphere-valued composite with any based map out of X+. The degenerate cases k=0 and N=∅ were treated in steps 1.1-2.1. Nothing beyond the inherited ACω is used: all maps are the canonical ones induced by φ and the supplied metrics.

DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6.1-sol)Open item page →

The Pontryagin-Thom map of a framed submanifold

Definition

Assume ACω (The Axiom of Countable Choice (ACω)). Let X be closed and smooth and let (N,φ) be a closed framed codimension-k submanifold, k≥0 (Framings of a normal bundle). A compatible chart Φ for (ν(N),id), a supplied smooth metric and a sufficiently small radius give a collapse c:X+→Th⁡(ν(N)) as in Pontryagin–Thom collapse with specified normal data. Compose it with the framing homeomorphism and projection of A framing identifies the Thom target with a sphere smash product: X+→cTh⁡(ν(N))→ΦφN+∧Sk→pSk. Its based homotopy class is the Pontryagin–Thom class. The map is based at the disjoint point of X+; its restriction to X need not preserve any preselected point of X.

For k≥1 we use the following normalized smooth representative f(N,φ). Fix an orientation-preserving stereographic coordinate z:Sk∖{∞}→Rk with centre y0=z−1(0) and positive basis b0=(dzy0)−1(e1,…,ek). Write u=φs(v) in a compatible tube Φ(s,v), and choose r>0 so that the chart is defined on {∣u∣≤2r}; compactness of N gives such a radius. Using The standard smooth step function, put a(q)=1−σ ⁣(q−r2/43r2/4). Thus a=1 for q≤r2/4, a>0 for q<r2 and a=0 for q≥r2. Define f(N,φ)(Φ(s,v))=z−1 ⁣(ua(∣u∣2))(∣u∣<r), and send all other points to ∞. Near N the target coordinate is exactly u. Near ∣u∣=r, the coordinate at infinity obtained by inversion is a(∣u∣2)u/∣u∣2, which is smooth and extends by zero because a is smooth and vanishes identically for ∣u∣≥r. Hence the map is smooth everywhere and constant near the boundary of the larger tube. Its centre preimage is exactly N.

This representative has the preceding collapse class. Use the framing metric ∣v∣φ=∣φs(v)∣. In the disk model of Disk bundle, sphere bundle, and Thom space: the differential topology interface, the smooth profile above has normalized radius b(t)=ta(t2)2+t2(0≤t≤r), with b(0)=0, b(r)=1 and b(t)>0 for t>0. Interpolating b(t) with t/r gives continuous disk-valued radial maps which agree at the boundary and never acquire an extra centre preimage. Closed pasting gives a based homotopy to the ordinary collapse, as in Continuity and smooth local representatives of collapse. Different metrics are compared by the radial Thom homeomorphisms; the next lemma proves tube independence.

For k=0, N is clopen in X, and f(N,φ):X→S0 sends N to the nonbasepoint and X∖N to the basepoint. For N=∅ it is constant at the basepoint. The framing is fixed data; a reflected framing changes the map by the corresponding sphere reflection. Countable choice is inherited only from the compatible-chart and normal-bundle machinery.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)Open item page →

Tube independence of the Pontryagin-Thom map

Statement

Assume ACω. Let X be a closed smooth manifold and let (N,φ) be a closed framed codimension-k submanifold of X, k≥0. Pontryagin-Thom maps of (N,φ) built from any two compatible tubular charts, any two supplied smooth metrics on ν(N⊆X), and any two sufficiently small positive radii are based homotopic as maps X+→Sk (The Pontryagin-Thom map of a framed submanifold).

The framing is fixed data throughout: the auxiliary choices removed here are the chart, the metric and the radius, and the framing-induced homeomorphism Φφ is the same structure transported along the metric comparison. A bundle automorphism of the normal datum other than the identity changes Φφ and is a change of framed submanifold, not a change of tube data; the construction makes no claim of independence under such an automorphism.

Facts & Assumptions

Given: A closed framed codimension-k submanifold (N,φ) of the closed smooth manifold X, and two sets of tube data for the normal datum (ν(N⊆X),id): compatible tubular charts, metrics h1,h2 on ν(N⊆X) and sufficiently small positive radii ρ1,ρ2.

[F1]

The Pontryagin-Thom map is f(N,φ)=p∘Φφ∘c, where c is the collapse of the given tube data and p,Φφ are the based projection and the framing-induced homeomorphism (The Pontryagin-Thom map of a framed submanifold).

[F2]

The collapse is continuous and based; collapses made with any two compatible tubular charts, sufficiently small radii and supplied metrics represent the same based homotopy class after the canonical radial identification of the metric targets (Continuity and smooth local representatives of collapse, Collapse homotopy for a fixed normal identification).

[F3]

The framing-induced homeomorphism is natural in the framed data and independent of the metric used on ν(N⊆X) up to the canonical radial homeomorphism (A framing identifies the Thom target with a sphere smash product).

[F4]

Based homotopies compose with fixed based maps: if H:X+×I→Y is a based homotopy and q:Y→Z is a based continuous map, then q∘H is a based homotopy; and based homotopy is an equivalence relation on based maps (Homotopies of continuous maps, homotopies relative to a subspace, and path homotopies relative to the endpoints).

Proof

1.1F2given

(The two collapses and the metric comparison.) Let ci:X+→Th⁡hi(ν(N⊆X)) be the collapse built from the i-th tube data, i=1,2. By [F2] both are based continuous maps, and there is a based homotopy between c1 and r−1∘c2, where r:Th⁡h1(ν)→Th⁡h2(ν) is the canonical radial comparison of metrics.

2.1F1F3F4step 1.1

(Composing with the framing identification.) Let Φφ(i):Th⁡hi(ν)→N+∧Sk be the framing homeomorphisms, and qi:=p∘Φφ(i). By [F3], Φφ(2)=Φφ(1)∘r−1 as based maps, hence q2∘r=q1: the metric comparisons and the framing homeomorphisms cancel exactly. Therefore f2=q2∘c2=q1∘(r−1c2) and f1=q1∘c1, and composing the based homotopy of step 1.1 with the fixed based map q1 gives a based homotopy f1≃f2 by [F4].

3.1F1F2F4step 1.1step 2.1∎

(Conclusion.) Steps 1.1-2.1 show that any two Pontryagin-Thom maps built from compatible charts, metrics and radii are based homotopic, the framing being held fixed. The argument used only continuity, the tube-independence of the collapse and the exact metric compatibility of the framing homeomorphism; no choice beyond the inherited ACω occurs, and for N=∅ both sphere-valued maps are constant at the basepoint. For k=0, N is clopen in X, and both maps send N to the nonbasepoint of S0 and its complement to the basepoint.

DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6.1-sol)Open item page →

Collapse of a framed neat cobordism in X times I

Definition

Assume ACω (The Axiom of Countable Choice (ACω)). Let (W,ε,Ψ) be a framed cobordism in X×I, with the literal product ends and constant collar framings of Framed cobordism of framed submanifolds. Extend W past both ends by the cylinders N0×(−1,0] and N1×[1,2) to a closed boundaryless embedded submanifold W~ of M=X×(−1,2). The product collars make these extensions smooth; extend Ψ constantly on them. The boundary normal quotients are those of Neat submanifolds have boundary-adapted slice charts, or directly the quotients of these extended product tangent bundles.

A compatible tube A of W~ with datum (W~×Rk,Ψ−1) exists by Compatible tubular charts realize a prescribed normal identification, now applied in the boundaryless manifold M. We can make this tube a product on smaller end collars as follows. Choose compatible tubes Bi of Ni in X with datum (Ni×Rk,φi−1) and let B be their products with time. Embed M properly in Euclidean space by Every smooth manifold embeds in some finite-dimensional Euclidean space; The Euclidean tubular neighbourhood theorem supplies a smooth retraction R of a Euclidean neighbourhood onto that image (normal addition followed by projection). Let χ be a smooth function of the base time, equal to one near the ends and supported within the original product collars, constructed using The standard smooth step function. On a small tube define C(w,v)=R((1−χ(w))A(w,v)+χ(w)B(w,v)), reading A,B,C in this Euclidean embedding and using A where χ=0. Both maps fix w and induce Ψw−1 on the normal quotient; derivatives of χ multiply B(w,0)−A(w,0)=0. Thus dC is invertible along zero: it is the identity on TW~ and an isomorphism on the normal quotient. By The smooth inverse function theorem on manifolds, C is locally a diffeomorphism there. It is injective on a sufficiently small uniform tube over compact W: otherwise distinct pairs with fibre coordinates tending to zero and equal images have convergent base subsequences; their limiting base points coincide because C(w,0)=w, contradicting local injectivity near that zero vector. Shrinking once more keeps the middle part away from ∂(X×I); on the collars C=B preserves time exactly. Consequently its restriction Φ over W is a neat compatible tube, product on smaller end collars. This is a local proof of the product-tube assertion used in Freed's Theorem 3.7 and the proof of Theorem 3.9, printed pp.26–27; boundaryless tube existence alone would not supply it.

In these framed coordinates choose r>0 with Φ defined on W×{∣v∣≤2r}. For k≥1, let z and a be the stereographic coordinate and smooth radius profile of The Pontryagin-Thom map of a framed submanifold, and define the collapse of the framed cobordism by cW(Φ(w,v))=z−1 ⁣(va(∣v∣2))(∣v∣<r),cW=∞ elsewhere. The same inversion-coordinate calculation proves smoothness, including across the cutoff sphere. On the product collars it is the product of the normalized collapse of (Ni,φi) with time; thus its endpoint restrictions are Pontryagin–Thom maps with compatible induced tube data. Extend by the constant basepoint on {∗}×I to obtain a based homotopy X+×I→Sk. For k=0, W is clopen in X×I; take its characteristic map to S0, whose endpoint restrictions are the characteristic maps of Ni. For empty W the map is constant. All formulas use supplied data and only the inherited countable-choice hypothesis.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)Open item page →

Framed cobordant submanifolds have homotopic Pontryagin-Thom maps

Statement

Assume ACω. Let X be a closed smooth manifold and let (N0,φ0), (N1,φ1) be closed framed codimension-k submanifolds of X, k≥0. If they are framed cobordant (Framed cobordism of framed submanifolds), then their Pontryagin-Thom maps X→Sk (The Pontryagin-Thom map of a framed submanifold) are homotopic, indeed based homotopic as maps X+→Sk; a framed cobordism supplies an explicit homotopy X×I→Sk whose restrictions at the two ends are the two Pontryagin-Thom maps up to based homotopy.

Facts & Assumptions

Given: A framed cobordism (W,ε,Ψ) in X×I from (N0,φ0) to (N1,φ1).

[F1]

The collapse cW:X×I→Sk of the framed cobordism is continuous and based, and its restrictions to X×{0} and X×{1} are collapses of (N0,φ0) and (N1,φ1) computed with the induced boundary tube data, hence representatives of the corresponding Pontryagin-Thom classes (Collapse of a framed neat cobordism in X times I).

[F2]

Pontryagin-Thom maps of a fixed framed submanifold built from different compatible tube data, metrics and radii are based homotopic (Tube independence of the Pontryagin-Thom map).

[F3]

The Pontryagin-Thom map is the based map p∘Φφ∘c:X+→Sk built from the collapse and the framing-induced homeomorphism (The Pontryagin-Thom map of a framed submanifold).

[F4]

Homotopies of based maps can be concatenated, reversed and composed with continuous maps in the time variable, and reversed homotopies are homotopies (Homotopies of continuous maps, homotopies relative to a subspace, and path homotopies relative to the endpoints).

Proof

1.1F1F3F4given

(The collapse is a homotopy between the end maps.) Choose tube data for the normal datum of W in X×I as in [F1] and form the collapse cW:X×I→Sk. By [F1], cW is continuous and based, and its restrictions cW(⋅,0) and cW(⋅,1) are the Pontryagin-Thom maps of (N0,φ0) and (N1,φ1) computed with the induced boundary tube data. Reading cW as a based homotopy X+×I→Sk between those two end maps, [F4] turns it into a based homotopy between the end maps.

2.1F2F3F4step 1.1

(Replacing the induced tube data.) The induced boundary tube data are compatible tube data for Ni in X; by [F2] the Pontryagin-Thom map of Ni computed with them is based homotopic to the Pontryagin-Thom map of (Ni,φi) computed with any other compatible tube data, in particular with the data used to define f(Ni,φi). Concatenating these two based homotopies with the end maps of step 1.1 yields a based homotopy X+×I→Sk from f(N0,φ0) to f(N1,φ1), by [F4].

3.1F1F2F4step 1.1step 2.1∎

(Conclusion.) Step 2.1 exhibits the required based homotopy; ignoring basepoints gives the homotopy of maps X→Sk, and the explicit homotopy is the collapse cW together with the two tube-comparison homotopies at the ends. For empty ends the maps are constant at the basepoint. For k=0, each path t↦cW(x,t) in the discrete space S0 is constant, so the end characteristic maps agree. No choice beyond the inherited ACω is used.

DefinitionDefinition: Literature-sourcedProof: Not applicableOpen item page →

Framed regular preimages of a map to a sphere

Definition

Assume ACω (The Axiom of Countable Choice (ACω)). Fix the standard orientation of Sk and identify Sk=Dk/Sk−1 with the one-point compactification Rk∪{∞} of the zero section of the trivial rank-k bundle over a point, so a chosen centre y0 corresponds to 0 (Trivial Thom spaces as suspension smash products, Disk bundle, sphere bundle, and Thom space: the differential topology interface). Let X be a closed smooth manifold, let f:X→Sk be smooth and let y∈Sk be a regular value, with b=(b1,…,bk) a positively oriented basis of TySk (Smooth manifolds and their smooth charts, Transversality to a point is the regular-value condition); read b as the trivialization β:TySk→Rk sending bi to the i-th standard basis vector (Orientation of a finite-dimensional real vector space).

Then N=f−1(y) is a closed embedded submanifold of X of codimension k, and the differential of f factors through the normal quotient to a smooth bundle isomorphism ν(N⊆X)⟶(f∣N)∗TSk; composing it with β gives a framing f∗b:ν(N⊆X)⟶N×Rk. The pair (N,f∗b) is the framed regular preimage of (f,y,b).

Both assertions are the case E=Rk of Transverse preimages carry the pulled-back normal structure, applied locally in a target chart centred at y, whose differential at y is β, to the trivial rank-k bundle over the one-point base: its Thom space is Sk=Rk∪{∞}, its zero section is {0}, transversality of f to {0} is exactly regularity of y (Transverse smooth maps, Transversality to a point is the regular-value condition), and clauses (i)-(ii) of that proposition give the closed embedded submanifold and the specified isomorphism of the normal quotient with the pulled-back target fibre; inserting the positive basis β turns the target factor into Rk and is precisely the framing. The construction is independent of any auxiliary choice: only the derivative of f along N, the value y and the basis b enter, so no chart of Sk at y is chosen and the framing is the composite displayed above. Regular values are not assumed to exist a priori; they are dense by Regular values form a dense Gδ set, and for k≥1 the framed cobordism class is shown to be independent of the regular value and positive basis in the two following lemmas.

Rank k=0 and N=∅ are included: for k=0 the sphere is S0, the basis is empty, β is the unique map TyS0→R0, and N=f−1(y) is a clopen submanifold of X of dimension dim⁡X, carrying the unique rank-zero framing; for N=∅ the normal bundle and the framing are empty. Regular-value independence does not extend to k=0: a constant map from a point to S0 has the point and the empty set as its two regular fibres. They are not framed cobordant within the point, since a compact neat codimension-zero submanifold of I is clopen, and one containing 0 must also contain 1. The countable-choice hypothesis is inherited exactly from Transverse preimages carry the pulled-back normal structure; the definition itself selects nothing.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)Open item page →

Positively oriented bases of an oriented vector space are path-connected

Statement

Let V be a finite-dimensional real vector space with an orientation (Orientation of a finite-dimensional real vector space). The set of positively oriented bases of V, with the topology it inherits from the linear isomorphisms V→Rdim⁡V, is path-connected; indeed any two positively oriented bases are joined by a smooth path of positively oriented bases, equivalently GL+(k,R)={A∈GL(k,R):det⁡A>0} is smoothly path-connected for every k≥0 (Invertible matrices and the general linear group GL⁡n(F), Paths, path-connected spaces and path components). In particular, for the oriented vector space TySk with its standard orientation, any two positive bases at y can be joined by a continuous path of positive bases.

Facts & Assumptions

Given: An oriented finite-dimensional real vector space V of dimension k, and the group GL+(k,R) of invertible real k×k matrices of positive determinant.

[F1]

Fixing one positively oriented basis b0 of V, the map A↦A(b0) is a bijection from GL+(V) (invertible endomorphisms of positive determinant) onto the set of positively oriented bases of V, with inverse given by the coordinate matrix in the basis b0; the determinant of the coordinate matrix detects positivity of the orientation (Orientation of a finite-dimensional real vector space, Invertible matrices and the general linear group GL⁡n(F)).

[F2]

Every invertible matrix is a finite product of elementary matrices, of three types: interchanges Spq, row scalings Dp(c) with c≠0, and row additions Tpq(c); the identity is the empty product (Elementary matrices obtained by applying one elementary row operation to an identity matrix, Every invertible finite square real matrix is a finite product of elementary matrices).

[F3]

For distinct indices p,q and t∈[0,1], let Rpq(t) be the matrix that is the identity off the plane span⁡{ep,eq} and equals (cos⁡(πt/2)−sin⁡(πt/2)sin⁡(πt/2)cos⁡(πt/2)) in the ordered basis (ep,eq). Its determinant is cos⁡2(πt/2)+sin⁡2(πt/2)=1, so Rpq(t) is invertible for every t; its entries are smooth in t by The derivatives of sine and cosine are cosine and minus sine (repeated differentiation alternates sine and cosine); and Rpq(0)=I while Rpq(1) sends ep↦eq, eq↦−ep and fixes the other standard basis vectors, so Spq=Rpq(1) Dq(−1) (Parity and the Pythagorean identity for sine and cosine, Sine and cosine are 1-Lipschitz on R, Rectangular matrix multiplication and the identity matrix In, including zero-sized shapes).

[F5]

A path in a topological space is a continuous map from I; the set of positive bases carries the subspace topology transferred by the bijection of [F1], so a continuous family of matrices gives a continuous family of bases (Paths, path-connected spaces and path components).

[F6]

The standard smooth step function σ is smooth, equals 0 on (−∞,0] and 1 on [1,∞); hence for smooth paths γ0:I→X, γ1:I→X with γ0(1)=γ1(0) the formula γ(t)=γ0(σ(2t)) for t≤12 and γ(t)=γ1(σ(2t−1)) for t≥12 is a smooth path, with all positive-order derivatives vanishing at the junction (The standard smooth step function).

Proof

1.1F1F5given

(Reduction to matrices.) Fix a positively oriented basis b0 of V. By [F1] the map A↦A(b0) is a bijection GL+(V)→{positive bases} whose inverse sends a basis to its coordinate matrix; a family t↦b(t) of bases is continuous exactly when its matrix entries in b0 are continuous. Choosing coordinates in b0 identifies GL+(V) with GL+(k,R), so it suffices to prove that GL+(k,R) is path-connected.

2.1F2F3F4step 1.1

(Deforming a factorisation to a diagonal sign matrix.) Let A∈GL+(k,R) and, by [F2], write A=E1⋯Em with each Ei elementary. Replace each factor by a continuous path Ei(t), t∈[0,1], of invertible matrices with Ei(0)=Ei: for Ei=Tpq(c) use Tpq((1−t)c), for Ei=Dp(c) with c>0 use Dp((1−t)c+t), for Ei=Dp(c) with c<0 use Dp((1−t)c−t), and for Ei=Spq use Rpq(1−t)Dq(−1), which starts at Spq by [F3] and ends at Dq(−1); the endpoint Ei(1) is I, I, Dp(−1) or Dq(−1) respectively, all diagonal with entries ±1. Every Ei(t) is invertible: a transvection has determinant one, the scaling paths have a diagonal entry that is a convex combination of the two nonzero numbers c and 1 (respectively c and −1) and so never vanishes, and the fourth path is a product of invertible matrices. Define A(t):=E1(t)⋯Em(t). Then A(t) is invertible for every t, A(0)=A, and A(1)=Δ is a product of matrices each of which is I or some Dp(−1), hence a diagonal matrix with entries ±1. Since t↦det⁡A(t) is continuous, never zero by invertibility, and positive at t=0, [F4] gives det⁡A(t)>0 for all t: so A is joined to Δ by a path in GL+(k,R).

3.1F3F4F6step 1.1step 2.1

(From the diagonal sign matrix to the identity.) The diagonal matrix Δ has det⁡Δ=det⁡A(1)>0, so the number of its entries equal to −1 is even. Pair the indices p<q carrying −1; for each pair, the matrix that is −1 on span⁡{ep,eq} and +1 elsewhere is realised by the block (cos⁡(π(1−t))−sin⁡(π(1−t))sin⁡(π(1−t))cos⁡(π(1−t))) at parameter t, which equals −I2 on that plane at t=0, the identity at t=1, and has determinant 1 throughout by [F3]. Doing this independently on the finitely many disjoint pairs and leaving the remaining coordinates fixed gives a continuous path Δ(t) of invertible matrices with Δ(0)=Δ, Δ(1)=I. Because Δ(t) is orthogonal of determinant 1, this path lies in GL+(k,R); concatenating it with the smooth path of step 2.1 by the smooth reparametrisation of [F6] yields a smooth path in GL+(k,R) from A to I.

4.1F1F2F4F5F6step 1.1step 2.1step 3.1∎

(Conclusion.) Every A∈GL+(k,R) is joined to I by a smooth path, and reversing paths joins any two elements of GL+(k,R) smoothly; hence GL+(k,R) is smoothly path-connected. By step 1.1 the set of positively oriented bases of V is connected by smooth paths of positive bases; applied to the oriented vector space TySk it gives the asserted smooth path of positive bases at y. The case k=0 is the one-point space GL+(0,R)={I0}. Every ingredient is an explicit formula, and the factorisation is a fixed finite one produced by the elimination theorem, so no choice principle is used.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passOpen item page →

Homotopic maps with a common regular value have framed-cobordant preimages

Statement

Assume ACω (The Axiom of Countable Choice (ACω)). Let X be closed and smooth, and let f0,f1:X→Sk, k≥0, be smoothly homotopic. If y is a regular value of both and b is a fixed positive basis of TySk, their framed regular preimages are framed cobordant in X.

Facts & Assumptions

Given: A smooth homotopy F:X×I→Sk, a common regular value y of its ends, and a positive basis b with coordinate isomorphism β:TySk→Rk.

[F1]

A smooth time reparametrization, constant near both endpoints, can be built from The standard smooth step function.

[F2]

Under countable choice, a smooth map transverse to a closed submanifold near a closed set can be perturbed to a transverse map without changing it on a smaller neighbourhood of that set (Relative transversality preserves a map on a closed good region).

[F3]

Transversality to a point is regularity (Transversality to a point is the regular-value condition). The local fibre-coordinate argument for a transverse preimage, including boundary transversality, gives a neat submanifold and its specified normal quotient isomorphism (Transverse preimages carry the pulled-back normal structure, (i)–(iv)). Composing the normal differential with β gives the framing of Framed regular preimages of a map to a sphere.

[F4]

Literal product ends with framings constant over their collars are precisely the data of Framed cobordism of framed submanifolds.

Proof

1.1F1F3givenconstruct

Reparametrize F by a smooth ρ:I→I equal to zero on [0,δ] and one on [1−δ,1], where 0<δ<1/2. Extend the resulting homotopy to a smooth map F~:X×R→Sk by f0 for t<0 and f1 for t>1. Smoothness across the ends follows from the constant collars. On a neighbourhood of the closed set A=X×((−∞,0]∪[1,∞)) this map is transverse to {y}, since its spatial derivatives there are those of fi, surjective at their y-preimages.

2.1F2step 1.1choose

Apply [F2] in the boundaryless manifold X×R, with Z={y} and closed set A. Obtain a transverse smooth map G equal to F~ near A. Compactness of X supplies 0<ε<δ with G(x,t)=f0(x) for 0≤t<ε and G(x,t)=f1(x) for 1−ε<t≤1: a finite cover of each compact end slice by product neighbourhoods gives a positive minimum time width.

3.1F3step 2.1construct

Set W=(G∣X×I)−1(y). In a local chart at y with differential β, [F3] makes W a closed, hence compact, neat codimension-k submanifold, with normal framing Ψ=β∘dG‾. The end restriction is transverse because it equals fi. Step 2.1 gives the literal product ends fi−1(y)×Θi, and there dG is the pullback of dfi and annihilates the time direction. Thus Ψ is the pullback of fi∗b throughout each collar. No global diffeomorphism extending the chosen target chart is required.

4.1F2F3F4step 3.1∎

By [F4], (W,ε,Ψ) is the required framed cobordism. Empty preimages cause no exception; for k=0 the preimages are clopen and the normal framings are the unique rank-zero maps. Countable choice is inherited from [F2] and [F3].

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passOpen item page →

The framed preimage class is independent of regular value and positive basis

Statement

Assume ACω (The Axiom of Countable Choice (ACω)). Let X be a closed smooth manifold and let f:X→Sk be smooth, with k≥1.

(i) At a regular value y of f, two positive bases b,b′ of TySk give framed-cobordant framed preimages (f−1(y),f∗b) and (f−1(y),f∗b′).

(ii) If y,y′ are regular values of f with positive bases b,b′, then the framed preimages (f−1(y),f∗b) and (f−1(y′),f∗b′) are framed cobordant (Framed regular preimages of a map to a sphere, Framed cobordism of framed submanifolds).

Consequently the framed cobordism class of a regular preimage depends only on the smooth homotopy class of f: smoothly homotopic maps have framed-cobordant preimages for any choices of regular values and positive bases.

Facts & Assumptions

Given: A closed smooth manifold X, a smooth map f:X→Sk, regular values and positive bases as in (i) and (ii), and the Pontryagin manifolds (f−1(y),f∗b) of Framed regular preimages of a map to a sphere.

[F1]

Any two positive bases of an oriented vector space are joined by a smooth path of positive bases (Positively oriented bases of an oriented vector space are path-connected).

[F2]

The cylinder N×I⊆X×I with a framing that is the pullback of f∗b near t=0 and of f∗b′ near t=1 is a framed cobordism from (N,f∗b) to (N,f∗b′) (Framed cobordism of framed submanifolds); the standard smooth step function smooths the two junctions obtained by concatenating the constant path at b, a path of positive bases and the constant path at b′, so a smooth path of positive bases can be reparametrised to be constant near t=0 and t=1 (The standard smooth step function).

[F3]

If f0,f1:X→Sk are smoothly homotopic and y is a regular value of both with fixed positive basis b, then the framed preimages are framed cobordant (Homotopic maps with a common regular value have framed-cobordant preimages).

[F4]

Framed cobordism is an equivalence relation, so framed cobordisms can be concatenated (Framed cobordism is an equivalence relation).

[F5]

Rotations of Rk+1 have determinant one and form a group; the rotations in a coordinate plane give explicit smooth one-parameter families (Orthogonal and unitary operators form groups, and their determinants have modulus one).

[F6]

Two smooth maps f0,f1:X→Sk, k≥1, have a common regular value: their critical sets are closed by the local rank-minor condition and compact because X is compact; their critical-value images are therefore closed and null by Sard. Each regular-value set is consequently open and dense, and the intersection of these two open dense sets is nonempty (Morse-Sard for smooth manifolds, Regular values form a dense Gδ set).

Proof

1.1F1F2given

(Basis independence (i).) Let N=f−1(y) and let b,b′ be positive bases of TySk. The change-of-basis matrix carries b to b′ and has positive determinant, so [F1] gives a smooth path bt, t∈[0,1], of positive bases with b0=b, b1=b′; by the reparametrisation recorded in [F2] we may take bt smooth and constant near t=0 and t=1. On N×I⊆X×I the normal bundle is canonically pr⁡N∗ν(N⊆X), and the formula Ψ(x,t):=f∗bt(x) defines a smooth bundle isomorphism ν(N×I⊆X×I)→N×I×Rk which over the end collar t∈[0,ε) equals the pullback of f∗b and over t∈(1−ε,1] the pullback of f∗b′. Hence (N×I,ε,Ψ) is a framed cobordism from (N,f∗b) to (N,f∗b′) by [F2].

1.2F5given

(A rotation family and a homotopy.) Let y,y′∈Sk be regular values of f and choose an orthonormal pair u,v in Rk+1 spanning a plane containing y when y′=−y; when y′=y take the constant identity family; otherwise there is an explicit rotation R1, identity on the orthogonal complement of a two-plane and equal to a plane rotation there, with R1(y)=y′: rotate in the plane span⁡{y,y′} if y′≠±y, and by angle π in a plane containing y if y′=−y. Let Rt be the corresponding family of rotations through angle tθ, so that t↦Rt is smooth, R0=id, R1(y)=y′, and each Rt has determinant one by [F5]. Put F(x,t):=Rt(f(x)); this is a smooth homotopy from f0:=f to f1:=R1∘f. The value y′ is a regular value of f0=f by hypothesis, and of f1 because f1−1(y′)=f−1(R1−1y′)=f−1(y) and df1=dR1∘df is surjective there.

2.1F3step 1.2

(The cobordism from the homotopy.) Apply [F3] to the smooth homotopy F from f to R1∘f and the common regular value y′ with the positive basis b′: the framed preimages (f−1(y′),f∗b′) and ((R1f)−1(y′),(R1f)∗b′) are framed cobordant. The second preimage is f−1(y); and since dR1 is invertible and orientation-preserving, the basis b′′:=dR1−1(b′) is positive at y and (R1f)∗b′=f∗b′′ by the chain rule: the framing of the preimage of a composition is the framing of the inner map read through the invertible differential. Hence (f−1(y′),f∗b′) is framed cobordant to (f−1(y),f∗b′′).

3.1F4step 1.1step 2.1

(Independence of the regular value (ii).) By step 1.1 applied to the regular value y of f and the two positive bases b′′ and b, the framed preimages (f−1(y),f∗b′′) and (f−1(y),f∗b) are framed cobordant. Concatenating this cobordism with the one of step 2.1 by [F4] gives a framed cobordism from (f−1(y′),f∗b′) to (f−1(y),f∗b), which is (ii).

4.1F3F4F6step 1.1step 3.1

(Consequence for smooth homotopies.) Let f0,f1:X→Sk be smoothly homotopic via F, and let yi be regular values of fi with positive bases bi, i=0,1. By [F6] choose a common regular value p of f0 and f1; choose any positive basis c of TpSk. Applying [F3] to the homotopy F and the common regular value p gives a framed cobordism between (f0−1(p),f0∗c) and (f1−1(p),f1∗c); applying (ii) of step 3.1 to f0 gives one between (f0−1(y0),f0∗b0) and (f0−1(p),f0∗c), and applying it to f1 gives one between (f1−1(y1),f1∗b1) and (f1−1(p),f1∗c). Concatenating the three framed cobordisms by [F4] gives a framed cobordism between (f0−1(y0),f0∗b0) and (f1−1(y1),f1∗b1), as asserted.

5.1F1F2F3F4F6step 1.1step 1.2step 2.1step 3.1step 4.1∎

(Conclusion.) Steps 1.1 and 3.1 prove (i) and (ii); step 4.1 proves that smoothly homotopic maps with any choices of regular values and positive bases have framed-cobordant preimages, so the framed cobordism class of a regular preimage depends only on the smooth homotopy class of the map. Empty preimages are included. The hypothesis k≥1 is essential: for a constant map X→S0, the two regular-value preimages are X and ∅, which need not be framed cobordant. Only ACω, inherited from the cited suppliers, is used.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passOpen item page →

The regular preimage of the collapse recovers the original framed submanifold

Statement

Assume ACω. Let (N,φ) be a closed framed codimension-k submanifold of a closed smooth X, with k≥1. For the normalized smooth Pontryagin–Thom representative f of The Pontryagin-Thom map of a framed submanifold, its centre y0 is regular, f−1(y0)=N, and the basis b0 fixed there induces exactly φ. Thus the framed regular preimage is (N,φ) on the nose. Arbitrary unnormalized collapses represent the same homotopy class but need not induce this literal framing.

Facts & Assumptions

Given: (N,φ), a compatible tube Φ inducing the identity on the normal quotient, and the normalized smooth representative f.

[F1]

With u=φs(v), the target coordinate of f(Φ(s,v)) is u/a(∣u∣2) for ∣u∣<r, with a=1 near zero and positive before the cutoff; elsewhere the value is ∞ (The Pontryagin-Thom map of a framed submanifold).

[F2]

A compatible chart fixes N and induces the identity on its normal quotient (Pontryagin–Thom collapse with specified normal data). The preimage framing is the differential on that quotient followed by the coordinate isomorphism determined by the target basis (Framed regular preimages of a map to a sphere).

Proof

1.1F1givenalgebra

Since φs is invertible and a is positive on the finite-value region, u/a(∣u∣2)=0 precisely when v=0. The remaining points map to ∞≠y0. Hence f−1(y0)=N.

2.1F1F2step 1.1algebra∎

Near v=0 the target coordinate is exactly φs(v). Its vertical derivative is φs, and its derivative on TsN is zero. Compatibility of the tube identifies the vertical quotient with ν(N)s by the identity, so the normal derivative of f in the b0 coordinates is exactly φs. It is surjective, proving regularity and f∗b0=φ.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passOpen item page →

The collapse of a regular preimage is homotopic to the original map

Statement

Assume ACω, let X be closed and smooth and let k≥1. If f:X→Sk is smooth, y is regular and b is a positive basis at y, then the Pontryagin–Thom map of (f−1(y),f∗b) is smoothly homotopic to f.

When y=y0 and b=b0 are the fixed centre and basis of The Pontryagin-Thom map of a framed submanifold, the homotopy has the following local form. There are a compact tube U and a smooth f1 equal to f outside U and to the normalized collapse g near N=f−1(y0); f≃f1 is supported in U, and f1≃g is constant near N. This local assertion requires the displayed target normalization.

More generally, if f is continuous and smooth on a neighbourhood of f−1(y), with surjective derivative there, the same collapse-class conclusion holds under continuous homotopy.

Facts & Assumptions

Given: f,y,b as in the statement, with N=f−1(y) and φ=f∗b.

[F1]

The normalized collapse g has centre fibre N and exactly the framing φ; in framing coordinates u its target coordinate is u near zero (The Pontryagin-Thom map of a framed submanifold, The regular preimage of the collapse recovers the original framed submanifold).

[F2]

Compatible tubes exist and a framing supplies product fibre coordinates (The tubular neighbourhood theorem in a smooth ambient manifold, Framings of a normal bundle). The differential defining the preimage framing is Framed regular preimages of a map to a sphere; the local-smooth transverse-preimage version, including closedness of the fibre, is Transverse preimages carry the pulled-back normal structure.

[F3]

Smooth maps paste over an open cover (Smooth maps paste over an open cover). Smooth cutoffs and endpoint-flat time reparametrizations are supplied by The standard smooth step function.

[F4]

Positive bases give framed-cobordant preimages for k≥1 (The framed preimage class is independent of regular value and positive basis); framed cobordisms give homotopic collapses (Framed cobordant submanifolds have homotopic Pontryagin-Thom maps). Continuously homotopic smooth maps are smoothly homotopic under countable choice (Continuously homotopic smooth maps are smoothly homotopic).

Proof

1.1F1F2givenalgebra

First suppose y=y0, b=b0. Shrink a framed tube so both maps lie in the centre coordinate chart there; use u=φs(v) as the fibre coordinate. In these coordinates F(s,0)=G(s,0)=0 and DuF(s,0)=DuG(s,0)=I, with G(s,u)=u on a sufficiently small tube by [F1]. A finite cover of compact N bounds the second fibre derivatives of F; integrating the derivative along each fibre segment gives ∣F(s,u)−u∣≤C∣u∣2 uniformly. Choose c>0 small enough that Cc<1/2, the estimates hold for ∣u∣≤c, and G(s,u)=u there. Thus for 0<∣u∣≤c, F(s,u)⋅u≥∣u∣2−C∣u∣3>0 and G(s,u)⋅u=∣u∣2>0. This uses framing coordinates, not a false positivity inference about an arbitrary invertible matrix.

2.1F3step 1.1construct

Let λ(∣u∣2) be a smooth cutoff equal to one for ∣u∣≤c/2 and zero for ∣u∣≥c. In the target centre chart put Ft=(1−tλ)F+tλG and use f elsewhere. On the support both vectors have positive dot product with u when u≠0, so no extra centre preimage is created. The family equals f on a neighbourhood of the tube boundary, hence pastes smoothly. Its final map f1 agrees with g on V={∣u∣<c/2} and with f outside a compact tube U.

3.1F3step 2.1construct

On X∖N, both f1 and g avoid y0. In stereographic coordinates h:Sk∖{y0}→Rk interpolate h(f1) linearly to h(g). On V use the constant family f1=g. These formulas agree on V∖N, so they paste to a smooth homotopy constant near N. Endpoint-flat reparametrization makes its concatenation with step 2.1 smooth. If N=∅, take U=V=∅ and simply interpolate the two maps in the chart avoiding y0.

4.1F4step 1.1step 2.1step 3.1construct

For general y,b, choose a rotation R joined smoothly to the identity with R(y)=y0; the usual plane rotation handles nonantipodal points, the identity handles equality, and a π rotation handles antipodes. Put f′=Rf. Its fibre at y0 is N, and the pushed basis dRyb induces exactly φ. By [F4], changing that positive basis to b0 gives a framed cobordism from (N,φ) to the framed preimage (N,φ′) of (f′,y0,b0). Their collapses are homotopic. Steps 1.1–3.1 compare f′ to the collapse of (N,φ′), while the rotation path compares f to f′. The concatenation proves the claimed homotopy; [F4] makes it smooth.

5.1F1F2F3F4step 1.1step 2.1step 3.1step 4.1∎

If f is only continuous away from its regular fibre, the local deformation of steps 1.1–2.1 is still smooth on a small tube and continuous elsewhere, and the chart interpolation of step 3.1 is continuous. For step 4.1, basis independence is the explicit framed cylinder using a path of positive bases, which requires only the local differential on the fibre. The same constructions therefore give a continuous homotopy to the collapse. The regular fibre is compact because it is closed in X. All choice is inherited from the stated suppliers.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)Open item page →

Based and free homotopy classes of maps between spheres agree

Statement

For n,k≥1 the forgetful map πn(Sk)→[Sn,Sk] from based to free homotopy classes is bijective. The spherical model of πn is identified with its cubical model by Cubical and spherical models of higher homotopy agree.

Facts & Assumptions

Given: Unit spheres Sn,Sk with fixed basepoints, n,k≥1.

[F1]

Orthogonal matrices form a group; determinants have modulus one (Orthogonal and unitary operators form groups, and their determinants have modulus one). Plane rotations have determinant one and can be continuously varied from the identity.

[F2]

Homotopies are continuous maps on the product, and a based homotopy fixes the basepoint (Homotopies of continuous maps, homotopies relative to a subspace, and path homotopies relative to the endpoints).

[F3]

The spherical and cubical based homotopy models agree (Cubical and spherical models of higher homotopy agree).

Proof

1.1F1F2givenconstruct

A rotation in a plane containing f(∗) and the target basepoint takes f(∗) to that basepoint and is joined to the identity by varying its angle. If the two points coincide use the identity, and if antipodal choose any perpendicular unit vector, available since k+1≥2. Postcomposition gives a free homotopy from f to a based map. This proves surjectivity.

1.2F1F4constructalgebra

For unit vectors a,b with a⋅b>−1, put K=baT−abT and Q(a,b)=I+K+K2/(1+a⋅b). On the plane spanned by a,b, this is the rotation taking a to b; on its orthogonal complement it is the identity. Direct multiplication gives Q(a,b)a=b and Q(a,b)TQ(a,b)=I; its determinant is one and Q(a,a)=I. The formula is continuous even at a=b. For a continuous path p:I→Sk, choose a finite subdivision so that p(t)⋅p(tj)>−1 on each subinterval, using uniform continuity. Set P0=I and inductively Pt=Q(p(tj),p(t))Ptj. Then Pt is continuous in SO(k+1) and Ptp(0)=p(t).

2.1F1F2step 1.2construct

Suppose based f0,f1 are freely homotopic by H. Apply step 1.2 to p(t)=H(∗,t) and define H^(x,t)=Pt−1H(x,t). This is a based homotopy from f0 to P1−1f1. Since p(0)=p(1)=∗, the matrix P1 fixes the basepoint vector and restricts to an element of SO(k) on its perpendicular subspace. Every element of SO(k) is joined to the identity by plane rotations: successively rotate its first column to the first coordinate vector, then its second column within the perpendicular complement, continuing until the final one-dimensional block, which is +1 because the determinant is one. At each stage an antipodal column is handled by a rotation through π in a two-plane; for k=1 the group is already the identity. Reversing the finite sequence and varying the angles gives the required path in the stabilizer of ∗. Postcomposing f1 with that path joins P1−1f1 to f1 through based maps. Concatenation with H^ proves injectivity.

3.1F3step 1.1step 2.1∎

Surjectivity and injectivity prove the assertion, including n=k=1, where the final stabilizer is trivial. All selections are finite. Via [F3] this is the stated bijection for the cubical group πn(Sk).

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passOpen item page →

The Pontryagin-Thom correspondence in fixed codimension

Statement

Assume ACω (The Axiom of Countable Choice (ACω) is inherited from the transversality and approximation suppliers). For n≥k≥1 the collapse construction and the framed-regular-preimage construction define mutually inverse bijections between

Equivalently, they give a bijection with the set [Sn,Sk] of free homotopy classes of continuous maps (Based and free homotopy classes of maps between spheres agree). The statement includes the empty preimage and the case n=k, where the framed submanifolds are zero-dimensional.

Facts & Assumptions

Given: Integers n≥k≥1, the sphere Sn, and the framed cobordism relation on closed framed (n−k)-submanifolds of Sn.

[F1]

The Pontryagin-Thom map f(N,φ) of a framed submanifold is a based continuous map, smooth in the radial cutoff model of the construction, and it is the composite of the collapse with the framing homeomorphism and the based projection (The Pontryagin-Thom map of a framed submanifold).

[F2]

Framed-cobordant submanifolds have based homotopic Pontryagin-Thom maps (Framed cobordant submanifolds have homotopic Pontryagin-Thom maps).

[F3]

Every continuous map Sn→Sk is homotopic to a smooth map, and continuously homotopic smooth maps are smoothly homotopic (Every continuous map between smooth manifolds is homotopic to a smooth map, Continuously homotopic smooth maps are smoothly homotopic).

[F4]

A smooth map has regular values, they are dense, and at a regular value with any positive basis the framed regular preimage is defined (Morse-Sard for smooth manifolds, Regular values form a dense Gδ set, Framed regular preimages of a map to a sphere).

[F5]

Along a smooth homotopy whose endpoints have a common regular value and fixed positive basis, the framed preimages are framed cobordant; and both the framed preimage class and the homotopy class of the collapse are independent of the regular value, the positive basis and the smooth representative (Homotopic maps with a common regular value have framed-cobordant preimages, The framed preimage class is independent of regular value and positive basis).

[F6]

The framed regular preimage of the Pontryagin-Thom map of (N,φ), at its centre and the corresponding positive basis, is (N,φ) on the nose (The regular preimage of the collapse recovers the original framed submanifold), and the Pontryagin-Thom map of the framed preimage of a smooth map is smoothly homotopic to that map (The collapse of a regular preimage is homotopic to the original map).

[F7]

Based and free homotopy classes of maps Sn→Sk agree (Based and free homotopy classes of maps between spheres agree).

[F8]

Framed cobordism is an equivalence relation, so "framed cobordism class" is a set of framed submanifolds (Framed cobordism is an equivalence relation).

Proof

1.1F1F2F7F8

(The map Φ on classes.) First take the free homotopy class of f(N,φ)∣Sn, and use the inverse of the forgetful bijection [F7] to define Φ(N,φ)∈πn(Sk). The disjoint basepoint of (Sn)+ does not by itself make that restriction based at a preselected point of Sn. By [F2], framed-cobordant framed submanifolds have based homotopic Pontryagin-Thom maps, so Φ is constant on framed cobordism classes and induces a map Φ from framed cobordism classes to πn(Sk).

1.2F3F4F5

(The map Ψ on classes.) For a continuous map f:Sn→Sk, choose a smooth map f′ homotopic to it by [F3], a regular value y of f′ and a positive basis b of TySk (a positive basis exists since the orientation of TySk has two classes and one flips sign by negating a vector), and set Ψ(f):=[(f′−1(y),f∗′b)], the framed cobordism class of the framed regular preimage. This is well defined: any two smooth maps homotopic to f are smoothly homotopic by [F3], and [F5] gives framed cobordism of the resulting preimages for different smooth representatives, different regular values and different positive bases. Hence Ψ induces a map Ψ:πn(Sk)→{framed cobordism classes}, defined without choosing a representative of the class: for every representative and every admissible choice the value is the same.

2.1F1F6step 1.2

(Ψ∘Φ is the identity.) Let (N,φ) be a closed framed (n−k)-submanifold and f=f(N,φ) its Pontryagin-Thom map, which is smooth by [F1]. Its centre y0 is a regular value and the framed preimage of (f,y0,b) is (N,φ) on the nose by [F6], where b is the positive basis corresponding to the structure identification. Therefore one admissible choice in the definition of Ψ gives the class of (N,φ), and by the well-definedness proved in step 1.2 every admissible choice gives it: Ψ(Φ(N,φ))=[(N,φ)].

2.2F5F6step 1.2

(Φ∘Ψ is the identity.) Let [f]∈πn(Sk) and let (N,φ) be the framed preimage produced by Ψ from a smooth map f′ homotopic to f, a regular value y and a positive basis b. Then Φ(Ψ[f])=[f(N,φ)], and f(N,φ) is smoothly homotopic to f′ by [F6]; since f′ is homotopic to f, the classes agree: Φ(Ψ[f])=[f].

3.1F1F5F7F8step 1.1step 1.2step 2.1step 2.2∎

(Conclusion.) Steps 1.1-1.2 define the two maps, and steps 2.1-2.2 show that their composites are the identities on the two sets; hence they are mutually inverse bijections. Composing Ψ with the identification of based and free classes [F7] gives the corresponding bijection with [Sn,Sk]. The case n=k is included: the preimages are zero-dimensional, and the empty manifold is allowed as a framed submanifold and as a preimage. Only ACω, inherited through the transversality, approximation and normal-bundle suppliers, is used.

DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6.1-sol)Open item page →

Stabilized framed cobordism and the framed bordism group

Definition

Assume ACω (The Axiom of Countable Choice (ACω) is inherited from the framed-cobordism relation). For d≥0 and k≥1 let Cobd(k) be the set of framed cobordism classes of closed framed d-submanifolds of Sd+k (codimension k; the fixed-codimension theorem applies with n=d+k≥k, Framed cobordism of framed submanifolds, Framed cobordism is an equivalence relation).

The equatorial stabilization σ=σk:Cobd(k)→Cobd(k+1) sends the class of (N,φ) to the class of (i(N),νeq⊕φ), where i:Sd+k↪Sd+k+1 is the equatorial inclusion, with the new ambient coordinate placed first, and νeq is the outward unit normal of the chosen northern hemisphere along its equator (a trivialization of the normal line, equipped with the product metric). It is well defined: i(N) is a closed embedded d-submanifold of Sd+k+1, the framing νeq⊕φ, the equatorial normal prepended to φ, is a trivialization of ν(i(N)⊆Sd+k+1)≅i∗ν(N⊆Sd+k)⊕ε1 over the equatorial collar, and if W⊆Sd+k×I is a framed cobordism from (N0,φ0) to (N1,φ1) then (i×idI)(W)⊆Sd+k+1×I is a framed cobordism from (i(N0),νeq⊕φ0) to (i(N1),νeq⊕φ1): the equatorial product structure is preserved by i, the product ends map to product ends, and the framing maps to the framing with the constant normal field νeq prepended. Hence σ carries classes to classes.

The framed bordism group Ωdfr is the colimit of the sequence Cobd(1)→ σ Cobd(2)→ σ Cobd(3)→ σ ⋯ , realised as the quotient of the disjoint union ⨆k≥1Cobd(k) by the equivalence relation generated by s∼σk(s) for all s∈Cobd(k) and all k. Thus an element of Ωdfr is represented by a framed d-submanifold of some sphere Sd+k, two representatives being equal exactly when they become framed cobordant after finitely many equatorial stabilizations; a stable framed bordism class is such a stabilization class of embedded framed manifolds. This is distinct from a stable framing on a fixed manifold, which is a trivialization after adding trivial summands, considered up to homotopy and further stabilization.

At a common level k≥2, define the proposed sum of two representatives by transporting addition in πd+k(Sk) through the fixed-codimension bijection Φk: x+ky=Φk−1(Φk(x)+Φk(y)). The stable theorem below proves compatibility with stabilization, so this operation descends to the colimit and is an abelian group law with the empty manifold as zero. It also proves its geometric interpretation: put the two framed representatives, with their transported framings, into separate affine half-space charts and take their disjoint union. The placement is part of this description; a union of two arbitrary intersecting or linked embeddings cannot be justified by the abstract unoriented bordism group law. No choice of representatives is built into the resulting operation, since independence is proved through the classifying bijection. The Pontryagin-Thom correspondence in fixed codimension supplies the levelwise bijections; The stable Pontryagin-Thom theorem identifies framed bordism with stable stems ↗ supplies the well-definedness of this colimit operation.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6.1-sol)Open item page →

Stabilizing a framed submanifold suspends its Pontryagin-Thom map

Statement

Assume ACω. For d≥0, k≥1, equatorial stabilization of a closed framed d-submanifold (N,φ) of Sd+k satisfies PT(σ(N,φ))=E(PT(N,φ))∈πd+k+1(Sk+1). The new equatorial normal is prepended, agreeing with the new first smash coordinate in the sphere-prespectrum convention. Thus the levelwise bijections intertwine stabilization and suspension.

Facts & Assumptions

Given: (N,φ) as above, its equatorial stabilization, and the suspension map E.

[F1]

The stabilized framing prepends the chosen equatorial normal, and stabilization respects framed cobordism (Stabilized framed cobordism and the framed bordism group).

[F2]

The sphere-prespectrum bonding map uses S1∧Sk≅Sk+1 with the new coordinate first (Suspension and sphere prespectra, Stable stems of the sphere). Smash products and their coherence are Smash product of based spaces and Canonical associativity, symmetry, and unit maps for smash products.

[F3]

The normalized collapse is smooth, constant off a small tube, and has its given framing as centre differential (The Pontryagin-Thom map of a framed submanifold). A continuous sphere-valued map smooth near a regular fibre is homotopic to the collapse of that framed fibre (The collapse of a regular preimage is homotopic to the original map).

[F4]

The fixed-codimension correspondence and the based/free identification are The Pontryagin-Thom correspondence in fixed codimension and Based and free homotopy classes of maps between spheres agree.

Proof

1.1F1F3F4construct

Choose a point outside N, possible because a positive-codimension submanifold has empty interior. A plane-rotation path can move that point to the chosen sphere basepoint; transporting N and its framing along this path gives a framed cobordism (flatten the time at its ends). Thus we may choose a representative avoiding the basepoint. Take its tube small enough to avoid that point too. Its collapse f is then based on Sn itself, n=d+k, and constant on a neighbourhood of its basepoint, with f−1(y0)=N and centre differential φ.

2.1F1F2F3step 1.1construct

Write the sphere complements as Rn and Rk, choosing an equatorial stereographic chart of the next sphere in which the equator is {0}×Rn and the chosen new normal points in the positive first coordinate. The smash S1∧Sn is the one-point compactification of R×Rn: the quotient of the compact product of the two one-point compactifications has exactly this open complement of its collapsed wedge, and its neighbourhoods at the collapsed point have compact complements. Under this identification idS1∧f is F(t,x)=(t,z(f(x)))(f(x)≠∞),F(t,x)=∞(f(x)=∞). It is continuous by the smash quotient, represents E[f], and near its centre fibre it is smooth. Its centre preimage is exactly {0}×N, with normal differential (a,v)↦(a,φ(v)), the prepended framing of [F1]. This is a statement about the class, not an assertion that a radial stabilized tube collapse equals a suspension pointwise.

3.1F3F4step 2.1∎

By the continuous local-smooth version of [F3], F is homotopic to the collapse of its framed centre preimage, namely σ(N,φ). Hence its free class is PT(σ(N,φ)), and [F4] identifies the corresponding based classes. Since [F]=E[f], the desired identity follows. Both constructions respect classes, so it gives a map of directed systems.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passOpen item page →

The stable Pontryagin-Thom theorem identifies framed bordism with stable stems

Statement

Assume ACω. For every d≥0, the levelwise Pontryagin–Thom bijections intertwine equatorial stabilization and suspension and induce an isomorphism Ωdfr≅πds. The group operation on the left is disjoint union after placing the framed representatives in separate affine charts; this operation is independent of representatives and of orientation-preserving parameterizations in those charts. The right-hand operation is addition of stable homotopy classes.

Facts & Assumptions

Given: The colimit set Ωdfr of Stabilized framed cobordism and the framed bordism group and the stable stem πds.

[F1]

The levelwise bijections Φk:Cobd(k)→πd+k(Sk) are The Pontryagin-Thom correspondence in fixed codimension; they satisfy Φk+1σ=EΦk by Stabilizing a framed submanifold suspends its Pontryagin-Thom map. The right-hand colimit is Stable stems of the sphere, also The sphere prespectrum groups are the classical stable stems.

[F2]

Cubical concatenation is the group operation and corresponds to the oriented spherical pinch sum (Higher homotopy group by based cubes, Cubical and spherical models of higher homotopy agree, The wedge of a family of pointed spaces). For n≥2 the group is abelian (Higher homotopy classes form groups and are abelian above degree one); for n=k=1 degree identifies it with Z (Based sphere maps are classified by degree). The exponential is smooth with derivative itself (The exponential function is smooth and (exp⁡)′=exp⁡), and its inverse logarithm (The natural logarithm as the inverse of the exponential function) has derivative 1/x on the positive reals (The natural logarithm has derivative 1/x and equals the integral from 1 to x of 1/t); differentiating 1/x repeatedly makes the logit coordinates and the packing paths below smooth.

[F3]

A normalized collapse is smooth and its centre fibre has its original framing (The Pontryagin-Thom map of a framed submanifold, The regular preimage of the collapse recovers the original framed submanifold). A smooth map with a regular framed fibre is homotopic to that fibre's collapse (The collapse of a regular preimage is homotopic to the original map).

[F4]

A compact smooth track of framed embeddings, constant near its time endpoints, gives a framed cobordism when its normal quotient is framed with those end restrictions (Framed cobordism of framed submanifolds). Framed cobordism preserves the collapse class (Framed cobordant submanifolds have homotopic Pontryagin-Thom maps). Endpoint flattening uses The standard smooth step function.

Proof

1.1F1given

The commuting levelwise bijections [F1] induce a bijection of the colimit sets: representatives equal after finitely many stabilizations have equal images, and conversely equality of images at a later level gives equality there by injectivity of Φk. Every class on the right comes from a finite level and has a preimage there. Hence there is a bijection Φ:Ωdfr→πds.

1.2F2F3F4construct

Put n=d+k. Choose representatives avoiding the sphere basepoint, by rotating a point outside each positive-codimension submanifold to that basepoint and transporting its framing along an endpoint-flat rotation track. Such tracks are framed cobordisms by [F4]. Choose small tubes avoiding the basepoint; their normalized collapses fi:Sn→Sk are based and constant near it. In orientation-preserving stereographic coordinates Sn∖{∗}=Rn, fix the cube-to-sphere quotient homeomorphism whose interior coordinate formula is xj=log⁡(sj/(1−sj)). It extends continuously to the quotient since approach to any cube face makes ∣x∣→∞. It is smooth on the interior; no smoothness of an arbitrary quotient homeomorphism is assumed.

2.1F2F3step 1.2constructalgebra

Let p(x)=ex/(1+ex), and define embeddings e0,e1:Rn→Rn by replacing the first coordinate with log⁡(p/(2−p)) and log⁡((1+p)/(1−p)), respectively, leaving the other coordinates unchanged. These are orientation-preserving diffeomorphisms onto the negative and positive first-coordinate half-spaces. They are exactly the inverse branches of the cubical pinch in the coordinates of step 1.2. Thus the map G equal to fi∘ei−1 in the respective half-space and to the basepoint on the separating sphere represents [f0]+[f1]. The nonconstant support of each fi is compact in Rn; its image under ei is compact and stays away from the separating sphere. Therefore G is constant near that sphere and the sphere basepoint, and is smooth everywhere. Its centre fibre is the disjoint union e0(N0)⊔e1(N1), with transported framings φi∘dei‾−1.

3.1F4step 2.1constructalgebra

Each packing preserves the individual framed class. An explicit path from the identity to ei replaces p(x1) by (1−t/2)p(x1) for i=0 or (1−t/2)p(x1)+t/2 for i=1, then applies the logit coordinate. Its derivative is positive for all t, so it is a smooth path Eti of embeddings on Rn. Flatten time at its endpoints and track compact Ni. At (Eti(x),t) the normal quotient is identified with ν(Ni)x by [(w,a)]⟼[(dEti)x−1(w−a ∂tEti(x))]. Indeed the track tangent vectors are (dEti(v)+a∂tEti,a) for v∈TxNi, so this formula is well defined and is an isomorphism. Composing with φi gives a smooth framing with the original framing at the first product end and the transported framing at the second. Thus [F4] preserves each individual class. We do not take the union of these two tracks, which could intersect.

4.1F1F2F3step 2.1step 3.1

By [F3], the collapse of the packed union in step 2.1 has class [G]=Φk(x)+Φk(y). Hence its framed class is exactly Φk−1(Φk(x)+Φk(y)), by [F1]. This proves the geometric interpretation of the proposed operation, its independence of the representatives used, and its independence of packing choices that preserve their transported individual classes in the two charts. Changing the orientation-preserving parameterization of either half-space induces a based degree-+1 self-map on its one-point compactification, hence a map homotopic to the identity by the degree classification in [F2]. Precomposition therefore leaves the two summand classes, and the resulting sum, unchanged. Compatibility with stabilization follows from Φk+1σ=EΦk and additivity of E. Thus the operation descends to the colimit; any two elements may be compared at k≥2, where the right-hand groups are abelian by [F2].

5.1F1F2step 1.1step 4.1∎

The colimit bijection of step 1.1 is additive for this disjoint-union law and sends the empty manifold to zero. Transporting inverses from πds supplies an inverse for every framed class; associativity, commutativity and identity follow from the same bijection. It is therefore an isomorphism of abelian groups. The proof establishes the well-definedness required by Stabilized framed cobordism and the framed bordism group and uses only the inherited countable-choice hypothesis.

RemarkRemark: Literature-sourcedProof: Not applicablejudge pass (gpt-6.1-sol)Open item page →

Normal framings, stable normal framings and tangential framings are distinct data

Remark

Assume ACω, inherited from the normal-bundle suppliers. Three different data types occur in this pair and are not interchangeable. An actual normal framing of a closed embedded submanifold N⊆Sm is a smooth bundle isomorphism φ:ν(N⊆Sm)→N×Rk for the honest normal bundle, with k the codimension of N in Sm (Framings of a normal bundle); it is part of the data of a framed submanifold and the Pontryagin-Thom map depends on it. A stable normal framing is a trivialization of ν(N)⊕εj for some j, considered up to homotopy and adding further trivial summands; this is the sense in which the stable normal bundle of a compact manifold is independent of the chosen embedding (Stable normal bundle of a compact smooth manifold, Stable normal bundle is independent of the embedding). A stable tangential framing is a trivialization of TN⊕εj′ considered up to homotopy and further stabilization.

The passage between the actual and the stable notions loses information. An actual framing φ of ν(N) determines the stable normal framing of φ⊕id on ν(N)⊕εj for every j, and, once the splitting TN⊕ν(N)=TSm∣N of the restricted ambient tangent bundle and the stable trivialization TSm⊕ε1≅εm+1 of the sphere are fixed, it also determines a stable tangential framing of N, by adding trivial summands to both sides; conversely a stable normal framing and the same ambient data determine a stable tangential framing up to homotopy. What is not available is a converse at the level of actual data: a stable framing is an equivalence class under adding trivial summands, it does not single out a codimension k, a level sphere Sm or an embedding, and no actual framing of ν(N) is recovered from it without choosing a level and splitting off the added summands.

Consequently the left-hand side of the stable Pontryagin-Thom theorem is a colimit. The elements of Ωdfr are stabilization classes of actual normal framings, not framed submanifolds of a fixed sphere (Stabilized framed cobordism and the framed bordism group), and only at a sufficiently large level does a representative of a stable class become an actual framed submanifold, where the levelwise correspondence of The stable Pontryagin-Thom theorem identifies framed bordism with stable stems becomes available. Identifying the fixed-codimension and the stable statements without keeping track of that stabilization is precisely the error this remark rules out.

5 · Examples, counterexamples and false statements

None yet.

Sources