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.

Root-of-unity skein TQFT for 3-manifolds

1 · Prerequisites

2 · Logical flowchart

3 · Definitions, theorems and proofs

DefinitionDefinition: Literature-sourcedProof: Not applicableverified 2026-10-22 (gpt-6.1-sol)Open item page →

The BHMV root-of-unity skein parameter

Definition

Fix an even integer p≥4 and a primitive 2p-th root of unity A∈C. Put

ζ=A2,q=−ζ,r=p/2,δ=−A2−A−2=q+q−1.

The coefficient field after specialization is C, and the Kauffman bracket loop coefficient is δ in the framed-link relation [L⊔O]=δ[L]. Define the quantum integers by [0]=0, [1]=1, and [j+1]=δ[j]−[j−1] for j≥1. Since q2=A4 has order r≥2, these quantum integers also satisfy

[j]=qj−q−jq−q−1=(−1)j−1A2j−A−2jA2−A−2.

The BHMV simple-label convention is the finite set

Jp={0,1,…,r−2}.

The pivotal loop and coefficient conventions throughout this page are the displayed δ and C; no zero quantum integer is inverted.

LemmaStatement: Literature-sourcedProof: AI-adaptedjudge pass (gpt-6-luna)verified 2026-10-22 (gpt-6.1-sol)Open item page →

Jones–Wenzl projectors exist up to the first zero quantum integer

Statement

Let p≥4 be even, let A be a primitive 2p-th root of unity, and use the quantum integers and Temperley–Lieb loop value δ=−A2−A−2 specified above. Write r=p/2. The standard Jones–Wenzl recursion defines nonzero idempotents pn for 0≤n≤r−1; their unnormalized closure traces are [n+1]. In particular, pr−1 has trace zero and is negligible, while the next standard recursive formula for pr is undefined because its denominator is [r]=0. This statement concerns that recursion and does not assert that no differently constructed normalized idempotent can exist at rank r.

Facts & Assumptions

Given: An even integer p≥4, a primitive 2p-th root A, the BHMV loop value δ=−A2−A−2, and the standard recursive idempotents.

[F1]

The BHMV root-of-unity skein parameter: q=−A2, q2=A4 has order r=p/2, [j]=(qj−q−j)/(q−q−1), and δ=q+q−1.

[F2]

Jones–Wenzl recursion over unit quantum integers: the recursive element pn is defined only when every denominator [2],…,[n] is a unit; the recursion for pn+1 uses [n+1]−1.

[F3]

Jones–Wenzl projectors: annihilation, uniqueness, partial trace, and closure: if [2],…,[n] are units, then pn is a nonzero idempotent, pnTLnpn=Cpn, and its unnormalized closure trace is [n+1].

[F4]

Diagram closure trace is cyclic and satisfies the tower rules: Tr⁡n(xy)=Tr⁡n(yx), including when δ=0.

[F5]

Negligible morphisms in the Temperley–Lieb category: an endomorphism f:n→n is negligible when Tr⁡n(g∘f)=0 for every g:n→n.

Proof

Given: The data and hypotheses in the Statement.

1.1givenF1algebra

Since q2 has exact order r, for 1≤j<r one has q2j≠1. Also q−q−1≠0, because q2≠1. Thus qj−q−j≠0 and [j]≠0 for every 1≤j<r. At j=r, q2r=1, so qr=q−r and [r]=0.

2.1givenF1F2F3step 1.1cases

For p≥6, one has r≥3, so every denominator [2],…,[r−1] is nonzero and therefore invertible in C. By [F2] and [F3], the recursion defines nonzero idempotents through pr−1 and gives Tr⁡r−1(pr−1)=[r]=0. For p=4, r=2 and pr−1=p1=11 is the initial projector; its closure is δ=[2]=0, so the same conclusion holds without using a rank-two denominator.

3.1F3F4F5step 2.1algebra

For any g∈TLr−1, [F3] gives pr−1gpr−1=λpr−1 for some λ∈C. Cyclicity [F4] gives Tr⁡(gpr−1)=Tr⁡(pr−1gpr−1)=λTr⁡(pr−1)=0. Thus pr−1 is negligible by [F5].

4.1F1F2step 1.1algebra∎

The recursive formula for pr requires the inverse of [r], which does not exist because [r]=0. Therefore that formula has no value at this step; in particular, assigning a value by cancelling a zero denominator is not part of the recursion. This leaves open the separate existence question for normalized idempotents constructed by other methods.

DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6-luna)verified 2026-10-22 (gpt-6.1-sol)Open item page →

Semisimplified Temperley–Lieb quotient at the BHMV root

Definition

Fix an even integer p≥4, a primitive 2p-th root A∈C, r=p/2, and loop value δ=−A2−A−2. Let Tδ be the complex Temperley–Lieb diagram category with objects the nonnegative integers and morphisms finite linear combinations of perfect matchings of bottom and top boundary points with no intersecting arcs. Compose by stacking and remove each closed component with factor δ. Define Nδ(m,n) to be the subspace of morphisms f:m→n such that Tr⁡m(g∘f)=0 for every reverse morphism g:n→m, where Tr⁡ is the unnormalized diagram-closure trace. Form the quotient category with the same objects and morphism spaces Hom⁡Tδ(m,n)/Nδ(m,n). Its additive envelope adjoins finite formal direct sums, with matrix morphisms and matrix composition. Its idempotent completion adjoins objects (X,e) for idempotents e∈End⁡(X) with e2=e, and morphisms (X,e)→(Y,f) are morphisms a:X→Y satisfying a=fae. Juxtaposition, planar cup-cap duality, and the induced trace descend to this completed quotient. This definition makes no assertion that it is semisimple, that its simple objects have a specified label set, or that any braid, twist, or modularity identities hold.

LemmaStatement: Literature-sourcedProof: AI-adaptedjudge pass (gpt-6-luna)verified 2026-10-22 (gpt-6.1-sol)Open item page →

Root Jones–Wenzl recursion and its negligible ideal

Statement

Fix even p≥4, a primitive 2p-th root A, and r=p/2. In the Temperley–Lieb diagram category over C with loop value δ=−A2−A−2, call f:m→n negligible when Tr⁡m(g∘f)=0 for every reverse morphism g:n→m, where Tr⁡ is the unnormalized diagram-closure trace. A tensor ideal is a family of linear subspaces of the morphism spaces closed under composition on either side and tensor product on either side by arbitrary morphisms. The negligible morphisms form a tensor ideal. This ideal is generated, as a tensor ideal, by the Jones–Wenzl projector pr−1, defined by the standard recursion through its first zero quantum integer; in particular the quotient has precisely the surviving projector labels 0,…,r−2.

Facts & Assumptions

Given: The specialized Temperley–Lieb category, its unnormalized closure trace, and the root-order projectors.

[F1]

Jones–Wenzl projectors exist up to the first zero quantum integer: pr−1 is a nonzero idempotent with zero closure trace.

[F2]

Temperley–Lieb planar diagram data and endomorphism modules: each morphism is a finite C-linear combination of noncrossing matching diagrams; cups and caps bend strands, and composition is stacking with each closed component weighted by δ.

[F3]

Negligible morphisms in the Temperley–Lieb category: f:m→n is negligible exactly when Tr⁡m(g∘f)=0 for every g:n→m.

[F4]

Negligible Temperley–Lieb morphisms form a tensor ideal: the negligible subspaces form a two-sided tensor ideal in the planar Temperley–Lieb category for every complex loop value.

[F5]

Diagram closure trace is cyclic and satisfies the tower rules: the unnormalized closure trace is cyclic and its identities do not divide by δ.

[F6]
[F7]

The rank-one tilting quotient matches the Temperley-Lieb ribbon convention: at the canonical root with order parameter r≥3, the negative-loop quotient has kernel equal to the embedded ideal generated by pr−1 and its ribbon quotient is equivalent to the specified rank-one tilting quotient.

[F8]

The root negligible quotient is semisimple with truncated fusion: for the positive-loop root q=eπi/r, the boundary projector ideal is the exact negligible ideal and the quotient has simple labels 0,…,r−2 with the truncated fusion rule.

Proof

Given: The data in the Statement.

1.1F1F2F3F5F6givenalgebra

By the boundary-projector result, pr−1 is an idempotent with trace zero. The diagram category [F2] supplies the endomorphism algebra TLr−1 and its closure trace. For any g∈TLr−1, [F6] gives pr−1gpr−1=λpr−1 for some λ∈C. Cyclicity [F5] gives Tr⁡r−1(g∘pr−1)=Tr⁡r−1(pr−1gpr−1)=λTr⁡r−1(pr−1)=0. Thus pr−1 itself is negligible by [F3].

2.1F3F4step 1.1algebra

By [F4], the negligible morphisms form a tensor ideal. Since [F3] and step 1.1 place pr−1 in that ideal, the tensor ideal it generates is contained in the negligible ideal.

2.2F1F7F8step 1.1algebra

Put A0=eπi/(2r). For r≥3, A0 is a primitive 4r-th root, q0=A02=eπi/r, and −A02−A0−2=−(q0+q0−1). Thus the negative-loop category in [F7] is exactly the BHMV Temperley–Lieb category at A0; its quotient kernel is the ideal generated by pr−1, and [F8] gives the complete label list.

2.3F1F3F4step 1.1algebracases

If r=2, the loop value is zero and p1=11 has zero trace by step 1.1. The tensor ideal it generates contains every identity 1n for n≥1, while the object 0 has endomorphism algebra C with trace 1. Hence the quotient is equivalent to finite-dimensional vector spaces with the single simple label 0; its negligible ideal is exactly the ideal generated by p1.

3.1F1F2step 2.2step 2.3algebra∎

For arbitrary primitive A, write A=A0k with gcd⁡(k,4r)=1. The cyclotomic automorphism σk of Q(A0) sends A0 to A, sends δ(A0) to δ(A), and sends every coefficient of each recursively defined projector and every closure-trace pairing matrix to its specialized coefficient at A. For each fixed pair of object labels, the matching set is finite by [F2], so the Hom spaces and trace-pairing matrices are finite-dimensional. The field automorphism preserves the ranks and kernels of all these matrices, the exact boundary ideal, and the quotient's simple labels. This completes the canonical-to-arbitrary-root transport and the proof.

Remarks

The canonical-root quotient and its boundary ideal use the full-rank path decomposition and the exact diagram-to-tilting comparison. Both supplier statements and their complete local proofs were inspected. The proof is transported to other primitive roots by the finite cyclotomic-field argument in step 3.1; no source-only acceptance is asserted.

LemmaStatement: Literature-sourcedProof: AI-adaptedjudge pass (gpt-6-luna)verified 2026-10-22 (gpt-6.1-sol)Open item page →

Root fusion rules and completeness of the simple objects

Statement

Let p≥4 be even, A∈C a primitive 2p-th root, and r=p/2. In the BHMV Temperley–Lieb quotient with loop value δ=−A2−A−2 and planar pivotal trace, the complete simple-label set is {0,1,…,r−2}, represented by the surviving Jones–Wenzl projectors pa obtained from the standard recursion using unit denominators [2],…,[a]. Every simple is self-dual. Its fusion multiplicities are

Nabc=1⟺∣a−b∣≤c≤min⁡(a+b, 2r−4−a−b) and a+b+c is even,

and are zero otherwise. With the BHMV overcrossing map ci=A1i+A−1Ui, where Ui is the cup-cap generator, the full-framing ribbon structure has planar quantum dimensions da=[a+1] and simple-object twists θa=(−1)aAa(a+2). Here the quantum integers use the fixed BHMV recurrence. The unnormalized Hopf-link matrix, whose (a,b) entry is the pivotal trace of the double braiding on pa⊗pb, is nondegenerate. Here modular means that this finite semisimple ribbon category has an invertible unnormalized Hopf-link matrix.

Facts & Assumptions

Given: The fixed BHMV parameter, its negligible quotient, and the standard planar pivotal convention.

[F1]

The BHMV root-of-unity skein parameter: r=p/2, δ=−A2−A−2, and [j] is defined by [0]=0, [1]=1, [j+1]=δ[j]−[j−1].

[F2]

Root Jones–Wenzl recursion and its negligible ideal: the quotient's negligible ideal is generated by pr−1, and its complete simple labels are 0,…,r−2.

[F3]

The rank-one tilting quotient matches the Temperley-Lieb ribbon convention: for r≥3 and q0=eπi/r, the negative-loop diagram quotient at loop −(q0+q0−1) with its diagram crossing and parity-corrected pivotal structure is ribbon-equivalent to the positive-pivotal rank-one tilting quotient. The diagram algebra is identified with the positive-loop algebra by Ui↦−ei.

[F4]

The rank-one tilting quotient is the truncated fusion category: the positive-pivotal quotient is semisimple, its simple objects are self-dual, and its tensor products have the truncated multiplicities.

[F5]

The standard root-of-unity sl2 tilting ribbon quotient is modular: the positive-pivotal quotient at q0=eπi/r has dimensions da+=sin⁡((a+1)π/r)/sin⁡(π/r), twists θa+=eπia(a+2)/(2r), and invertible sine Hopf-link matrix.

[F6]

Modular ribbon fusion category and Hopf-link data: modularity means that the unnormalized Hopf-link matrix of the finite semisimple ribbon category is invertible.

[F7]

Semisimplified Temperley–Lieb quotient at the BHMV root: each fixed pair of object labels starts from a finite set of matching diagrams, and the quotient morphism space is a quotient of that finite-dimensional span.

Proof

Given: The root parameter and quotient in the Statement.

1.1F1F2givencasesalgebra

If r=2, then p=4, δ=0, and the sole proposed retained label is 0. The boundary projector is p1=11 and has zero trace. The tensor ideal it generates contains 1n=11⊗1n−1 for every n≥1. Therefore every morphism with a positive source or target is zero in the quotient, because it factors through the corresponding identity. The endomorphism space of object 0 is C, and its trace is the identity, so it is non-negligible. The quotient's additive idempotent completion is thus finite-dimensional vector spaces with one self-dual simple 0. Its fusion is 0⊗0=0, its unit twist is 1, and its Hopf-link matrix is the 1×1 matrix (1), which is invertible.

1.2F2F3F4F5given

Suppose r≥3 and first take the canonical root A0=eπi/(2r), so q0=A02=eπi/r and the BHMV loop is −(q0+q0−1). By [F3], the diagram quotient with its parity-corrected pivotal structure is ribbon-equivalent to the positive-pivotal quotient of [F4]–[F5]. Thus the quotient has exactly the labels 0,…,r−2, the displayed truncated fusion multiplicities, self-dual simples, and a nondegenerate Hopf-link matrix for that corrected pivotal structure.

2.1F1F3F4F5F6step 1.2algebra

On the semisimple quotient define εpa=(−1)a 1pa. This is a monoidal natural automorphism: each retained summand pc of pa⊗pb has c≡a+b(mod2) by [F4], so εpc=εpaεpb; it is dual-compatible because every pa is self-dual. The planar pivotal structure is ε times the parity-corrected one of [F3]. Multiplying a pivotal structure j by a dual-compatible monoidal automorphism ε changes the twist u−1j to εθ, preserves balancing by monoidality, and preserves dual compatibility; hence the planar convention is also ribbon. Its dimensions are da=(−1)ada+, which equal the BHMV recurrence value [a+1]: the sequence vj=(−1)j−1[j]q0 has v0=0, v1=1 and satisfies vj+1=−(q0+q0−1)vj−vj−1, so uniqueness of the recurrence gives vj=[j] for the BHMV loop. Its simple twists are θa=(−1)aθa+=(−1)aA0a(a+2). Its unnormalized Hopf-link entries are sab=εaεbsab+, since the pivotal trace on pa⊗pb is multiplied by εaεb. Therefore its Hopf-link matrix is obtained from the invertible matrix in [F5] by multiplying row a and column b by (−1)a and (−1)b, so it remains invertible. The criterion in [F6] proves modularity.

3.1F1F7step 1.2step 2.1algebra∎

For any primitive A, write A=A0k with gcd⁡(k,4r)=1. The automorphism of the cyclotomic field Q(A0) sending A0 to A transports the finite diagram coefficients, the recursively defined projectors, the braiding and twist matrices, and the Hopf-link matrix. By [F7], every fixed Hom space is finite-dimensional, so this transport preserves all ranks and nonzero determinants. The fusion multiplicities are dimensions of these finite Hom spaces and so are unchanged under the field automorphism and extension to C. Thus the canonical-root result applies to every primitive 2p-th root with the exact stated BHMV convention; in particular θa=(−1)aAa(a+2).

DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6-luna)verified 2026-10-22 (gpt-6.1-sol)Open item page →

Weighted Lagrangian extension and root anomaly data

Definition

For the fixed root parameter, let Cp denote the semisimplified Temperley–Lieb category with labels 0,…,r−2, pivotal dimensions dj=[j+1], and Hopf-link data inherited from the Kauffman bracket. Put Dp2=∑j=0r−2dj2 and choose Dp∈C with that square. Choose κ∈C satisfying κ6=A−6−(p+1)p/2. The root-specific extended data are the weighted Lagrangian ribbon bordisms whose objects are closed oriented labeled surfaces with a Lagrangian subspace of real first homology and whose morphisms are collared oriented labeled ribbon bordisms with integer weights. If M1:Σ0→Σ1 and M2:Σ1→Σ2 have relations ΛM1,ΛM2 and boundary Lagrangians λ0,λ1,λ2, define

K1=ΛM1(λ0),K0=λ1,K2=ΛM2 t(λ2),n(M2∘M1)=n2+n1+μ(K1,K0,K2),

using the Wall-signature Maslov convention. The category composition, units, and anomaly weight are those specified by this formula and the weighted Lagrangian bordism definition; no ordinary oriented anomaly-free functor is part of this Definition.

Remarks

The BHMV source uses p1-structures, while the registered prerequisite map specifies Lagrangian-weighted data. This Definition supplies the latter using the existing local category carrier; the equality between its Maslov correction and the BHMV p1/signature anomaly remains a theorem obligation and is not asserted here.

LemmaStatement: Literature-sourcedProof: Literature-sourcedjudge pass (gpt-6-luna)verified 2026-10-22 (gpt-6.1-sol)Open item page →

Root Kirby-color handle-slide and blow-up invariance

Statement

Fix even p≥4, a primitive 2p-th root A, the BHMV semisimplified category, and its quantum dimensions dj and twists θj for 0≤j≤r−2. Put Ωp=∑j=0r−2djpj, Δ+=∑jdj2θj, and Δ−=∑jdj2θj−1. The Statement asserts that both Δ+ and Δ− are nonzero. For a finite integral framed link L in the oriented three-sphere, color each surgery component by Ωp and normalize the bracket evaluation by Δ+b+(L)Δ−b−(L), where b±(L) are the positive and negative indices of the framing matrix. With the stated BHMV root and weighted anomaly convention, the resulting value is unchanged under framed isotopy, handle slides, and addition or deletion of a split ±1-framed unknot.

Facts & Assumptions

Given: The root parameter, the semisimplified category, its simple projectors, quantum dimensions, twist, and a finite integral framed surgery link.

[F1]

The BHMV root-of-unity skein parameter fixes p, A, r, δ, and C.

[F2]

Semisimplified Temperley–Lieb quotient at the BHMV root defines the quotient and the root-specific projectors.

[F3]

Root fusion rules and completeness of the simple objects supplies the resolved finite simple list, dimensions, twists and invertible Hopf-link matrix in the registered root/pivotal convention. Its accepted interface is reused, not its body.

[F4]

Weighted Lagrangian extension and root anomaly data fixes the weighted Lagrangian composition and the selected sixth root κ.

[F5]

External input — actual BHMV Kirby color and Gauss units. BHMV, Topological Quantum Field Theories Derived from the Kauffman Bracket, Topology34(1995), §2 p.897, the displayed definition of θp and equations(*)–(**), together with §5.8 p.913: for the primitive 2p-th root and even p=2r, the Kirby element is Ωp=∑j=0r−2djpj and Δε=η−1κ3ε(ε=±1),η∈C×. Here κ is the selected BHMV SIXTH root from [F4], not the square-root weight parameter used by some later sources. The source's usual signed quantum dimensions are exactly dj=[j+1] for the registered negative-loop recurrence. Its recalled invariant θp(S3(L))=⟨L(Ωp)⟩/(Δ+b+Δ−b−) includes the actual Kirby handle-slide identity. This established identity is imported, not claimed as a new local recoupling proof.

[F6]

External input — weighted normalization and exact sign. BHMV §2 pp.897–898 and Gilmer–Masbaum, arXiv:0912.4706v3, §2 pp.4–6 and §11 pp.23–25, give the weighted version. Put ν=κ3; one weight step multiplies the map by ν. The closed TQFT normalization is Zp(M,w)=ηb0(M)+b1(M)νwθp(M), componentwise with the banded data. It is this additional normalization, not raw θp alone, that supplies the finite TQFT. Gilmer–Masbaum's Maslov form on (L1+L2)∩L3 is (a1+a2)⊙(b1+b2)=ω(a2,b1). Modulo its radical it is the negative of the registered Wall form ω(a,b′) with a+b+c=0. Thus their correction −μGM is precisely the registered +μWall. Their square-root parameter is ν, explicitly the cube of BHMV's parameter. No choice of an unrelated square root Dp is silently identified with η−1; only Dp2=η−2 follows.

Proof

Given: The data in the Statement.

1.1F1F2givenconstruct

The label set is finite, so Ωp is a finite linear combination of colored projectors. The normalized surgery expression is therefore a finite complex number whenever the displayed Gauss sums are nonzero.

2.1F1F2F3F5step 1.1algebra

By [F5], both Gauss sums are units, since η and κ are units. Framed isotopy preserves the ribbon bracket. The imported Kirby-color identity preserves its value under a handle slide, while the framing matrix changes by an integral congruence and hence has unchanged positive and negative indices. A split signed-unit unknot multiplies the numerator by Δε and increases exactly the corresponding index by one. The denominator therefore cancels that factor. Deletion is the inverse operation. This proves the displayed surgery invariance with the exact normalization of the Statement.

3.1F4F5F6step 2.1algebra∎

The weighted interpretation is [F6]: the signature unit is ν=κ3, with the source's Maslov sign converted to the registered Wall order. The distinct additional ηb0+b1 normalization is retained when these closed values enter the TQFT pairing; no anomaly-free ordinary-oriented functor follows. The finite color and cancellation calculations add no choice assumption.

DefinitionDefinition: Literature-sourcedProof: Not applicableverified 2026-10-22 (gpt-6.1-sol)Open item page →

Skein TQFT state space and handlebody pairing

Definition

Assume full AC for the surgery-presentation and universal-construction inputs below. Fix an even integer p≥4, a primitive 2p-th root A∈C, put r=p/2 and δ=−A2−A−2, and use the selected BHMV sixth root κ satisfying κ6=A−6−(p+1)p/2. Fix a closed oriented marked surface Σ with its Lagrangian and weighted anomaly data. Let IΣ be the finite union of its marked ribbon-end bands, with each band labeled by its specified simple object aj∈{0,…,r−2}. For connected Σ, fix marked handlebodies H,H′ whose boundaries are both identified orientation-preservingly with Σ, so that H∪Σ(−H′) is the standard oriented S3. Both filling modules carry the specified boundary labels before reversal; orientation reversal of the second filling dualizes its boundary labels and reverses its ribbon-end arcs. Disconnected and empty surfaces are included by the tensor and unit convention below.

For a banded boundary link I in ∂H, let KA(H;I) be the complex specialization of the relative Jones–Kauffman skein module generated by framed banded links in H meeting ∂H transversely in exactly I, modulo framed isotopy fixed on ∂H and the following local relations in interior balls. In a chart where the over-strand runs southwest to northeast and the under-strand northwest to southeast, let LA join the two left endpoints and the two right endpoints, and let LB join the top endpoints and the bottom endpoints, and require the crossing and both smoothings to carry untwisted blackboard framing, with no additional full twists and matching framed endpoint collars. Impose [L×]=A[LA]+A−1[LB]. If O is a zero-framed trivial circle bounding a disk disjoint from the rest of the link, impose [L⊔O]=δ[L]. The empty link is allowed.

Let I~Σ be the boundary link obtained by replacing each band Ij by aj parallel bands, deleting it when aj=0. On KA(H;I~Σ), the identity-coefficient Jones–Wenzl projector paj acts on the corresponding parallel strands in a boundary collar. Define the colored relative skein module by KA(H;IΣ,a)=PaKA(H;I~Σ), where Pa is the product of these mutually disjoint boundary projectors. Every denominator in these projectors is among [2],…,[aj], which are nonzero because aj≤r−2. When IΣ is empty, this is the ordinary closed-link module KA(H).

Let r:KA(H′;IΣ,a)→KA(−H′;−IΣ,a∗) be the semilinear orientation-reversal map: for a finite skein sum, r(∑LcL[L])=∑LcL‾[−L], with the second filling, its ribbon data and anomaly data reversed. The bar is complex conjugation, so A‾=A−1 and κ‾=κ−1. Reversal exchanges the two bracket smoothings, and conjugation exchanges their coefficients, so r respects the skein relations and the boundary projectors. For x∈KA(H;IΣ,a) and y∈KA(H′;IΣ,a), glue x to r(y) along the matching boundary bands and idempotents. Denote the resulting graph by Γ(x,r(y)). Define BΣ(x,y)=Zp(S3,Γ(x,r(y)),w) with the prescribed gluing weight. This is linear in x and conjugate-linear in y, as in BHMV Proposition 1.9, p.891. Ordinary gluing of two already oppositely oriented skein inputs is instead bilinear; the reversal here is part of the Definition. Define

Rad⁡(BΣ)={x∈KA(H;IΣ,a):BΣ(x,y)=0 for every y∈KA(H′;IΣ,a)},Vp(Σ)=KA(H;IΣ,a)/Rad⁡(BΣ).

This is the radical-quotient handlebody state-space construction for a connected marked surface. For a finite disjoint union Σ=⨆a=1kΣa with the direct-sum component reference Lagrangian, define Vp(Σ)=⨂a=1kVp(Σa), and define its pairing as the product of the component sesquilinear pairings on pure tensors. Set Vp(∅)=C and B∅(z,w)=zw‾. Permuting components uses the ordinary tensor symmetry; the empty tensor is the unit. These are the external BHMV conventions (Q1) p.885 and (M) p.886, whose validity for even p≥4 is part of Theorem 1.4 p.887; Gilmer–Masbaum §11 p.24 also states the empty-surface unit. For an arbitrary specified real Lagrangian, including one that does not split over components, use the weight-zero identity cylinder from the componentwise handlebody-kernel reference to that Lagrangian in the registered weighted theory. The reverse cylinder is its inverse since μ(L,L′,L)=0, exactly as in the existing real-reference transport for weighted gluing; this does not assume every real Lagrangian splits or is rational. The tensor conventions are taken at the reference objects and then transported by these inverse cylinders. Independence of the marking, finite generation, nondegeneracy and the gluing maps remain separate mathematical obligations supplied below by the stated external BHMV/weighted theory inputs.

External normalization input. In this Definition the closed marked evaluation is the BHMV TQFT normalization, not merely the raw signature-normalized closed invariant θp of the Kirby lemma. BHMV §2 pp.897–898 and Gilmer–Masbaum §11 pp.23–25 give Zp(M,Γ,w)=ηb0(M)+b1(M)(κ3)wθp(M,Γ),Δ±=η−1κ±3. Let gp=12∑m=12p(−1)mAm2, and let the bar denote the BHMV involution A↦A−1, κ↦κ−1. The normalizing scalar is η=(Aκ)3(A2−A−2)p−1gp‾∈C×. Thus Zp(S3,∅,0)=η and Zp(S2×S1,∅,0)=1. These nonzero normalizations and the registered Wall-sign weight are part of the pairing. BHMV Proposition 1.9, p. 891, applies to a general prescribed banded boundary link I: the map from the relative skein module to the universal state space is surjective and its kernel is the left radical of the complementary pairing. For colored boundary bands, BHMV §§4.5–4.8 realize each color by the corresponding parallel cable and Jones–Wenzl projector. BHMV Theorem 4.11, p. 909, states that admissibly colored relative spines whose boundary is the prescribed colored link form an orthogonal basis. The selected κ remains a sixth root, with one weight step acting by its cube. The source is used at these located statements and constructions, not as a claimed locally reconstructed universal-construction proof.

Source hypotheses for this identification. The normalized BHMV closed invariant is multiplicative and involutive, so BHMV Proposition 1.1 gives its cobordism-generated quantization functor at this fixed root; extend its coefficients to C as stated above. BHMV Theorem 1.7 supplies both hypotheses needed for Proposition 1.9: its Kauffman bracket relations let that proposition replace the free relative bordism module by the specialized relative skein module, and its surgery axioms give the surjectivity and radical-kernel description. BHMV's colored expansion in §§4.5–4.8 transports the same construction to the boundary projectors defining KA(H;IΣ,a). The corresponding weighted Lagrangian version is the BHMV skein TQFT with integral weights and Lagrangians in place of p1-structures, as described by Gilmer–Masbaum in §11; their one-weight multiplier is matched to the selected BHMV sixth root by κGM=κ3. In that section Lemma 11.1 states the weighted surgery axiom and Remark 11.3 applies it to the handlebody skein presentation; the source refers the Lemma 11.1 proof to the BHMV argument rather than repeating it. These are imported source inputs, not a locally reconstructed proof, and they do not add another quotient or change the state-space interface.

Definition references

LemmaStatement: Literature-sourcedProof: Literature-sourcedverified 2026-10-22 (gpt-6.1-sol)Open item page →

Finite generation by admissible handlebody-spine states

Statement

Let (H,IΣ,a) be a marked handlebody with prescribed boundary bands IΣ and boundary labels a. Fix a finite banded relative spine G⊂H, meaning a banded graph whose degree-one boundary vertices give exactly IΣ, whose interior vertices have degree two or three, and whose regular neighborhood is H; for the unmarked 3-ball, the empty graph is allowed. Let each edge carry a label in {0,…,r−2}, with every boundary half-edge on a band Ij labeled by its prescribed color aj. Call a labeling admissible when labels agree across every interior 2-valent vertex and the three labels at each interior trivalent vertex satisfy the root fusion rule. In the radical-quotient state space Vp(Σ), the states represented by admissibly colored relative spines span, and the set of such labelings is finite. This statement asserts spanning, not linear independence.

Facts & Assumptions

Given: A finite marked handlebody spine and the root state-space construction.

[F1]

Skein TQFT state space and handlebody pairing defines the state space as a radical quotient of the specialized colored relative handlebody skein module with its prescribed boundary bands.

[F2]

Root fusion rules and completeness of the simple objects proves that the finite label set is {0,…,r−2} and gives the exact local fusion rule; it does not itself prove that arbitrary handlebody skeins are spanned by spine states.

[F3]

External input — actual relative handlebody quotient and spine basis. BHMV, Proposition 1.9 p. 891 and Theorem 4.11 p. 909, identify the radical-quotient relative handlebody model with the universal state space and give an orthogonal basis represented by the p-admissibly colored fixed banded spine whose boundary is the prescribed colored banded link. For even p=2r, the source's q=(p−2)/2=r−1 and colors <q are exactly 0,…,r−2; p-admissibility is the registered truncated fusion rule. Coefficients are specialized through the given complex primitive root, with no vanishing quantum denominator inverted. This imports the actual spanning theorem, rather than inferring it from abstract semisimplicity alone.

Proof

Given: The fixed finite relative spine G, its prescribed boundary labels, and the finite label set {0,…,r−2}.

1.1givenF2construct

The edge set of G is finite and each edge has only r−1 candidate labels by [F2]. The prescribed boundary colors are fixed, so there are at most (r−1)∣E(G)∣ edge labelings; the admissible ones form a finite subset. If G is empty, its unique labeling is the empty labeling.

2.1F1F2F3step 1.1∎

Apply [F3] to the supplied relative spine, using [F1] for the matching colored relative radical-quotient model and [F2] for its root labels and vertex rule. Every vector is a linear combination of the admissible relative-spine states. Together with the finite count in step 1.1 this proves the full Statement. Linear independence, although present in the imported theorem, is not needed for this spanning conclusion. No infinite skein reduction or additional choice campaign is performed.

LemmaStatement: Literature-sourcedProof: Literature-sourcedverified 2026-10-22 (gpt-6.1-sol)Open item page →

Handlebody pairing nondegeneracy and gluing

Statement

For the fixed root parameter, the normalized BHMV sesquilinear pairing between the relative skein modules of two complementary marked handlebodies induces a perfect pairing between their radical quotients, using the coefficient-conjugate space on the oppositely oriented side. The inverse pairing defines the sewing map for gluing along a closed marked surface, and these maps satisfy the stated weighted Lagrangian anomaly correction.

Facts & Assumptions

Given: The finite spanning set of admissible spine states and the normalized gluing pairing.

[F1]

Skein TQFT state space and handlebody pairing defines the radical quotient and pairing.

[F2]

Finite generation by admissible handlebody-spine states supplies the finite admissible spine spanning set via its explicitly labelled external BHMV input.

[F3]

Weighted Lagrangian extension and root anomaly data fixes the weight and Maslov convention.

[F4]

External input — perfect pairing and sewing. BHMV, Theorem1.4 p.887, property(F) on p.886, Proposition1.9 p.891, and Theorem4.11 p.909: in this even-root theory the radical quotient is finite free with unimodular dual pairing, the admissible spine states form a basis with unit diagonal norms in the double pairing, and the universal gluing maps satisfy the TQFT axioms. Proposition1.9 permits the complementary handlebody giving S3, rather than assuming the self-double and S3 pairings have the same matrix. Their normalization is the Zp normalization explicitly fixed in [F1]. Only perfectness/basis/sewing at this actual convention is imported.

[F5]

External input — actual weighted variant. Gilmer–Masbaum, arXiv:0912.4706v3, §2 equation(1) pp.4–6 and §11 pp.23–25, states the colored TQFT with Lagrangians and integer weights and one-weight multiplier ν=κ3 in BHMV notation. The finite sign comparison μGM=−μWall is as recorded in the Kirby lemma, so its composition is exactly that of [F3]. The source states rational Lagrangians. For the real Lagrangians of [F3], take the rational handlebody-kernel reference on each marked surface and transport by the weight-zero identity cylinder to its specified real Lagrangian; the reverse cylinder is its inverse because μ(L,L′,L)=0. Conjugating morphisms by these cylinders is a functor to the rational-reference category, with the integer weights computed by [F3], so the imported functor extends to these real objects. This is a finite categorical transport, not an assumption that every real Lagrangian is rational.

Proof

Given: A marked handlebody spine, its admissible colorings, and the gluing pairing.

1.1F1F2F3given

On a fixed finite spine, each admissible coloring gives a finite colored graph state. Glue the first state to the semilinear orientation reversal of the second state, as in [F1], and evaluate the resulting closed colored graph using the specified root bracket and anomaly normalization.

2.1F1F2F4step 1.1algebra

By [F4], the pairing descends to a perfect sesquilinear pairing V×W→C, where V is the quotient from H and W is the quotient from the second filling H′ with the same oriented boundary, by the right radical. The semilinear reversal in [F1] identifies W‾ linearly with the quotient for −H′. Thus B(v,w‾):=BΣ(v,w) is a perfect bilinear pairing between V and W‾. This is stronger than left nondegeneracy by the definition of a radical alone. Choose bases vi and w‾j, and let Bij be their bilinear pairing matrix. Its inverse gives the coevaluation tensor ∑i,j(B−1)jivi⊗w‾j; contraction with B on either side is the identity. This finite linear-algebra identity is exactly the sewing contraction in the imported gluing theorem and is independent of the chosen dual bases.

3.1F3F4F5step 2.1algebra∎

The imported weighted gluing in [F5] gives the prescribed anomaly factor, with weight law [F3]. Its parameter is the cube of the selected sixth root, not an independently chosen square root with an unchecked sign. The real-Lagrangian transport in [F5] respects composition because the intervening reference cylinders cancel. Thus the inverse-pairing sewing maps satisfy precisely the registered weighted law. No unweighted anomaly cancellation or positivity of every complex-root Hermitian form is asserted.

TheoremStatement: Literature-sourcedProof: Literature-sourcedverified 2026-10-22 (gpt-6.1-sol)Open item page →

Root-of-unity skein TQFT gluing and surgery invariance

Statement

Assume full AC. Fix even p≥4, a primitive 2p-th root A, and the BHMV Kauffman bracket convention with δ=−A2−A−2. On the weighted Lagrangian extended bordism category built from the root semisimplified Temperley–Lieb data, the normalized Kirby-color surgery maps define a finite-dimensional skein TQFT: the handlebody state spaces are the radical quotients of their skein modules, gluing is given by the inverse normalized handlebody pairing, and the resulting closed value is unchanged under isotopy, framed handle slides, and split ±1-blow-ups. Forgetting the weight and Lagrangian extension yields only the projective ordinary-oriented data unless a separate anomaly-cancellation result is proved.

Facts & Assumptions

Given: Full AC, the fixed BHMV root and pivotal convention, weighted Lagrangian bordism data, the root state spaces and their surgery maps.

[A1]

The Axiom of Choice supplies the full choice assumption used by the surgery-presentation and Kirby-move supplier.

[F1]

Kirby moves classify integral framed-link presentations: framed surgery diagrams for the same closed oriented manifold are connected by the stated isotopies, framed handle slides, and split signed-unit-unknot moves.

[F2]

Root Kirby-color handle-slide and blow-up invariance supplies the closed surgery invariance, nonzero Gauss sums and precise external BHMV weighted normalization, including weight unit κ3.

[F3]

Handlebody pairing nondegeneracy and gluing supplies the perfect radical-quotient pairing, finite inverse sewing tensor and externally specified weighted gluing.

[F4]

Weighted Lagrangian ribbon bordisms supplies associative weighted Lagrangian composition with the Wall-signature Maslov correction.

Proof

Given: The data in the Statement.

1.1A1F2F3F4givenconstruct

For a fixed surgery presentation, the color set is finite and the state-space pairing is defined by a finite bracket evaluation. This constructs a candidate linear map for each presented weighted bordism.

2.1A1F1F2F3F4step 1.1

Any two presented closed surgery links are related by [F1], and [F2] preserves their values under every move. For bordisms, use the BHMV additional ηb0+b1 normalization specified in the state-space Definition, not the raw closed invariant alone. The finite state spaces, perfect pairing and actual surgery/sewing maps are [F3], whose external BHMV/Gilmer–Masbaum input supplies the TQFT construction at the exact even root and complex specialization. The inverse-pairing contraction composes these maps, with the integer Wall/Maslov weight law [F4]. This proves finite-dimensionality, gluing and presentation invariance of the stated weighted theory.

3.1A1F2F3F4step 2.1∎

Forgetting the weight removes the scalar that corrects the Maslov/signature defect. Therefore the imported construction descends only projectively to ordinary oriented data without a further anomaly-cancellation input. No such additional theorem is asserted here. Full AC remains the declared branch of the presentation/marked category interfaces; finite color sums add no stronger assumption.

4 · Examples, counterexamples and false statements

None yet.

Sources