Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Eqs. (12)–(15) — rung eigenvalue identity

Proved
QuantumSixJ.Network.rung_eigenvalue

by igor.kh · Nov 13, 2026 · Mathlib 0df444a (Lean v4.33.1)

monoidal-categoriesquantum-groupsspin-networks

Let C\mathcal CC be a KKK-linear preadditive monoidal category, K=C(q)K=\mathbb C(q)K=C(q), with additive and KKK-linear tensor product. The chosen objects VnV_nVn​ for natural labels satisfy V0≅1V_0\cong\mathbb1V0​≅1 and have exact self-pairings. Assume SuQ2Axioms: unique scalar endomorphisms of each VnV_nVn​, zero morphisms between different labels, the admissible SL(2)\mathrm{SL}(2)SL(2) fusion decompositions, and the loop values Δn\Delta_nΔn​. Choose trivalent vertices WWW, and assume VertexNormalized W: theta normalization and cyclicity. Scalars act through ι:Q(q)↪K\iota:\mathbb Q(q)\hookrightarrow Kι:Q(q)↪K, fixing qqq. Composition f;gf;gf;g means first fff, then ggg.

Also assume the separate initial data TetSeed W: for every natural a,b,c,d,ja,b,c,d,ja,b,c,d,j, T(a,b,c,d;j,2)=ι(Tet⁡sum(a,b,c,d;j,2)) id1T(a,b,c,d;j,2)=\iota(\operatorname{Tet}^{\mathrm{sum}}(a,b,c,d;j,2))\,\mathrm{id}_{\mathbb1}T(a,b,c,d;j,2)=ι(Tetsum(a,b,c,d;j,2))id1​. These are seed values, not an all-channel network-to-sum identity.

For K2, the rung operator weighted by [a][b][a][b][a][b], and Kop, its extension to the first two of four legs, prove three identities for all natural labels:

  1. Closing the triangle gives exactly
ηl;((split⁡l;ab;K2ab;fuse⁡ab;l)⊗idVl);ϵl=ι([a][b]) T(a,a,b,b;l,2).\eta_l;((\operatorname{split}_{l;ab};K2_{ab};\operatorname{fuse}_{ab;l})\otimes\mathrm{id}_{V_l});\epsilon_l =\iota([a][b])\,T(a,a,b,b;l,2).ηl​;((splitl;ab​;K2ab​;fuseab;l​)⊗idVl​​);ϵl​=ι([a][b])T(a,a,b,b;l,2).

The cup/cap and strand order are those of the specified exact self-pairing. 2. split⁡l;ab;K2ab=ι(Λ(a,b,l))split⁡l;ab\operatorname{split}_{l;ab};K2_{ab}=\iota(\Lambda(a,b,l))\operatorname{split}_{l;ab}splitl;ab​;K2ab​=ι(Λ(a,b,l))splitl;ab​. 3. X(a,b,c,d;l);Kopabcd=ι(Λ(a,b,l))X(a,b,c,d;l)X(a,b,c,d;l);\mathrm{Kop}_{abcd}=\iota(\Lambda(a,b,l))X(a,b,c,d;l)X(a,b,c,d;l);Kopabcd​=ι(Λ(a,b,l))X(a,b,c,d;l).

No admissibility restriction is imposed: invalid-channel cases must follow from the category and zero conventions. Neither a triangle evaluation nor an eigenvector equation is added as a hypothesis.

Independent mathematical read-back

Rung closure and eigenvalue identities

Setting and hypotheses

Let F=Q(q)F=\mathbb Q(q)F=Q(q), K=C(q)K=\mathbb C(q)K=C(q), and let ι:F→K\iota:F\to Kι:F→K be the coefficient embedding fixing the indeterminate qqq. Work in an arbitrary category C\mathcal CC, with arbitrary object and morphism universes, that is preadditive and KKK-linear, monoidal, monoidally preadditive, and monoidally KKK-linear. Thus composition and tensoring respect the indicated additive and scalar structures. No braiding, symmetry, dagger, positivity, or concrete representation realization is specified.

The label data consist of objects VnV_nVn​ for every natural number nnn, an isomorphism V0≅1V_0\cong\mathbf1V0​≅1, and a chosen exact self-pairing on every VnV_nVn​. Write its cup and cap as Dn:1→Vn⊗VnD_n:\mathbf1\to V_n\otimes V_nDn​:1→Vn​⊗Vn​ and Cn:Vn⊗Vn→1C_n:V_n\otimes V_n\to\mathbf1Cn​:Vn​⊗Vn​→1; exactness includes both snake identities. All ordered tensor factors below retain their displayed order. Composition f;gf;gf;g means first fff, then ggg. Write f⋆gf\mathbin{\star}gf⋆g for composition with the canonical associators and unitors inserted; it does not permute strands. Right and left whiskering are denoted f▹Af\triangleright Af▹A and A◃fA\triangleleft fA◃f.

The following are assumed category axioms, not conclusions:

  1. For every nnn and every endomorphism fff of VnV_nVn​, there is a unique s∈Ks\in Ks∈K such that f=s idVnf=s\,\mathrm{id}_{V_n}f=sidVn​​.
  2. Every map Va→VbV_a\to V_bVa​→Vb​ is zero when a≠ba\ne ba=b.
  3. For every a,ba,ba,b, there exist maps ic:Vc→Va⊗Vbi_c:V_c\to V_a\otimes V_bic​:Vc​→Va​⊗Vb​ and pc:Va⊗Vb→Vcp_c:V_a\otimes V_b\to V_cpc​:Va​⊗Vb​→Vc​, indexed by all natural ccc, such that ic;pc=idVci_c;p_c=\mathrm{id}_{V_c}ic​;pc​=idVc​​ for admissible c≤a+bc\le a+bc≤a+b, and ∑cpc;ic=idVa⊗Vb\sum_c p_c;i_c=\mathrm{id}_{V_a\otimes V_b}∑c​pc​;ic​=idVa​⊗Vb​​, summing over precisely those channels. These existential maps are not stipulated to equal the chosen vertex maps below.
  4. Dn;Cn=ι(Δ(n)) id1D_n;C_n=\iota(\Delta(n))\,\mathrm{id}_{\mathbf1}Dn​;Cn​=ι(Δ(n))id1​.

A triple of integers (a,b,c)(a,b,c)(a,b,c) is admissible exactly when all three entries are nonnegative, all three triangle inequalities hold, and a+b+ca+b+ca+b+c is even. Put

[n]=qn−q−nq−q−1,[n]!={0n<0,∏r=1n[r]n≥0,Δ(n)={(−1)n[n+1]n≥0,0n<0.[n]=\frac{q^n-q^{-n}}{q-q^{-1}},\qquad [n]!=\begin{cases}0&n<0,\\\prod_{r=1}^n[r]&n\ge0,\end{cases} \qquad \Delta(n)=\begin{cases}(-1)^n[n+1]&n\ge0,\\0&n<0.\end{cases}[n]=q−q−1qn−q−n​,[n]!={0∏r=1n​[r]​n<0,n≥0,​Δ(n)={(−1)n[n+1]0​n≥0,n<0.​

For admissible (a,b,c)(a,b,c)(a,b,c), with s=(a+b+c)/2s=(a+b+c)/2s=(a+b+c)/2, define

θ(a,b,c)=(−1)s[s+1]![s−a]![s−b]![s−c]![a]![b]![c]!;\theta(a,b,c)=(-1)^s\frac{[s+1]![s-a]![s-b]![s-c]!}{[a]![b]![c]!};θ(a,b,c)=(−1)s[a]![b]![c]![s+1]![s−a]![s−b]![s−c]!​;

otherwise it is zero.

Choose arbitrary trivalent covectors tabc:Va⊗Vb⊗Vc→1t_{abc}:V_a\otimes V_b\otimes V_c\to\mathbf1tabc​:Va​⊗Vb​⊗Vc​→1. The following formulas specify all bent vertices exactly, including reversals of labels. Identity maps at the ends of these formulas are retained to indicate canonical regrouping:

tabc=Dc⋆(Vc◃(Db▹Vc))⋆(Vc◃((Vb◃(Da▹Vb))▹Vc))⋆(tcba▹(Va⊗Vb⊗Vc))⋆id,slab=id⋆(Da▹Vl)⋆(Va◃((Db▹Va)▹Vl))⋆(Va◃Vb◃tbal)⋆id,fabl=id⋆((Va⊗Vb)◃Dl)⋆(tabl▹Vl)⋆id.\begin{aligned} t^{abc}={}&D_c\star(V_c\triangleleft(D_b\triangleright V_c)) \star\bigl(V_c\triangleleft((V_b\triangleleft(D_a\triangleright V_b))\triangleright V_c)\bigr) \star(t_{cba}\triangleright(V_a\otimes V_b\otimes V_c))\star\mathrm{id},\\ s_l^{ab}={}&\mathrm{id}\star(D_a\triangleright V_l) \star\bigl(V_a\triangleleft((D_b\triangleright V_a)\triangleright V_l)\bigr) \star(V_a\triangleleft V_b\triangleleft t_{bal})\star\mathrm{id},\\ f_{ab}^{l}={}&\mathrm{id}\star((V_a\otimes V_b)\triangleleft D_l) \star(t_{abl}\triangleright V_l)\star\mathrm{id}. \end{aligned}tabc=slab​=fabl​=​Dc​⋆(Vc​◃(Db​▹Vc​))⋆(Vc​◃((Vb​◃(Da​▹Vb​))▹Vc​))⋆(tcba​▹(Va​⊗Vb​⊗Vc​))⋆id,id⋆(Da​▹Vl​)⋆(Va​◃((Db​▹Va​)▹Vl​))⋆(Va​◃Vb​◃tbal​)⋆id,id⋆((Va​⊗Vb​)◃Dl​)⋆(tabl​▹Vl​)⋆id.​

Here tabct^{abc}tabc has outputs a,b,ca,b,ca,b,c, slab:Vl→Va⊗Vbs_l^{ab}:V_l\to V_a\otimes V_bslab​:Vl​→Va​⊗Vb​, and fabl:Va⊗Vb→Vlf_{ab}^{l}:V_a\otimes V_b\to V_lfabl​:Va​⊗Vb​→Vl​. The superscripted vertex is a linear cup-bending construction, not a Hermitian adjoint.

The vertex-normalization hypothesis has exactly two fields:

tabc;tabc=ι(θ(a,b,c)) id1,t^{abc};t_{abc}=\iota(\theta(a,b,c))\,\mathrm{id}_{\mathbf1},tabc;tabc​=ι(θ(a,b,c))id1​,

and cyclicity R(tabc)=tbcaR(t_{abc})=t_{bca}R(tabc​)=tbca​, where the exact rotation is

R(tabc)=id⋆(Da▹(Vb⊗Vc⊗Va))⋆(Va◃(tabc▹Va))⋆Ca.R(t_{abc})=\mathrm{id}\star(D_a\triangleright(V_b\otimes V_c\otimes V_a)) \star\bigl(V_a\triangleleft(t_{abc}\triangleright V_a)\bigr)\star C_a.R(tabc​)=id⋆(Da​▹(Vb​⊗Vc​⊗Va​))⋆(Va​◃(tabc​▹Va​))⋆Ca​.

There is also a separate seed hypothesis. Define

Xabcd;l=Dl⋆(slab⊗slcd)⋆id,Habcd;lR=id⋆(Va◃(fbcl▹Vd))⋆tald.X_{abcd;l}=D_l\star(s_l^{ab}\otimes s_l^{cd})\star\mathrm{id},\qquad H^R_{abcd;l}=\mathrm{id}\star\bigl(V_a\triangleleft(f_{bc}^{l}\triangleright V_d)\bigr)\star t_{ald}.Xabcd;l​=Dl​⋆(slab​⊗slcd​)⋆id,Habcd;lR​=id⋆(Va​◃(fbcl​▹Vd​))⋆tald​.

For signed channels j,lj,lj,l, the graphical tetrahedron is

T(a,b,c,d;j,l)={Xadcb;j;Hadcb;lRj,l≥0,0otherwise.T(a,b,c,d;j,l)= \begin{cases}X_{adcb;j};H^R_{adcb;l}&j,l\ge0,\\0&\text{otherwise}.\end{cases}T(a,b,c,d;j,l)={Xadcb;j​;Hadcb;lR​0​j,l≥0,otherwise.​

The boundary order is exactly a,d,c,ba,d,c,ba,d,c,b, the first channel jjj labels the cup in XXX, and the last channel lll is the inner horizontal rung. The seed assumes, for all natural a,b,c,d,ja,b,c,d,ja,b,c,d,j,

T(a,b,c,d;j,2)=ι(T(a,b,c,d;j,2)) id1.T(a,b,c,d;j,2)=\iota(\mathcal T(a,b,c,d;j,2))\,\mathrm{id}_{\mathbf1}.T(a,b,c,d;j,2)=ι(T(a,b,c,d;j,2))id1​.

It is additional initial-value data, not a field of vertex normalization, and not a conclusion that arbitrary channels agree with a scalar sum.

For completeness the seed's scalar T\mathcal TT is independently defined as follows. Set

A=(a+d+j2,b+c+j2,a+b+l2,c+d+l2),B=(b+d+j+l2,a+c+j+l2,a+b+c+d2).A=\left(\frac{a+d+j}{2},\frac{b+c+j}{2},\frac{a+b+l}{2},\frac{c+d+l}{2}\right),\quad B=\left(\frac{b+d+j+l}{2},\frac{a+c+j+l}{2},\frac{a+b+c+d}{2}\right).A=(2a+d+j​,2b+c+j​,2a+b+l​,2c+d+l​),B=(2b+d+j+l​,2a+c+j+l​,2a+b+c+d​).

If (a,d,j),(b,c,j),(a,b,l),(c,d,l)(a,d,j),(b,c,j),(a,b,l),(c,d,l)(a,d,j),(b,c,j),(a,b,l),(c,d,l) are all admissible, then

T(a,b,c,d;j,l)=∏i=14∏r=13[Br−Ai]![a]![b]![c]![d]![j]![l]!∑z=0a+b+c+d+j+l(−1)z[z+1]!∏i=14[z−Ai]!∏r=13[Br−z]!;\mathcal T(a,b,c,d;j,l)= \frac{\prod_{i=1}^4\prod_{r=1}^3[B_r-A_i]!}{[a]![b]![c]![d]![j]![l]!} \sum_{z=0}^{a+b+c+d+j+l} \frac{(-1)^z[z+1]!}{\prod_{i=1}^4[z-A_i]!\prod_{r=1}^3[B_r-z]!};T(a,b,c,d;j,l)=[a]![b]![c]![d]![j]![l]!∏i=14​∏r=13​[Br​−Ai​]!​z=0∑a+b+c+d+j+l​∏i=14​[z−Ai​]!∏r=13​[Br​−z]!(−1)z[z+1]!​;

otherwise it is zero. Integer halves use integer division; admissibility makes the displayed halves integral in the nonzero branch.

The three asserted conclusions

Define the two-leg operator and its four-leg extension by

Kab=ι([a][b]) ((saa2▹Vb)⋆(Va◃f2bb)),K^abcd=id⋆(Kab▹(Vc⊗Vd))⋆id.K_{ab}=\iota([a][b])\, \bigl((s_a^{a2}\triangleright V_b)\star(V_a\triangleleft f_{2b}^{b})\bigr),\qquad \widehat K_{abcd}=\mathrm{id}\star(K_{ab}\triangleright(V_c\otimes V_d))\star\mathrm{id}.Kab​=ι([a][b])((saa2​▹Vb​)⋆(Va​◃f2bb​)),Kabcd​=id⋆(Kab​▹(Vc​⊗Vd​))⋆id.

Thus the label-2 rung is split from the aaa strand and fused into the bbb strand, with exactly the prefactor [a][b][a][b][a][b], not a divided or square-root prefactor.

For admissible (a,b,l)(a,b,l)(a,b,l), put

λab(l)=[(a−b+l)/2][(−a+b+l)/2]−[(a+b−l)/2][(a+b+l)/2+2][2],\lambda_{ab}(l)= \frac{[(a-b+l)/2][(-a+b+l)/2]-[(a+b-l)/2][(a+b+l)/2+2]}{[2]},λab​(l)=[2][(a−b+l)/2][(−a+b+l)/2]−[(a+b−l)/2][(a+b+l)/2+2]​,

and put λab(l)=0\lambda_{ab}(l)=0λab​(l)=0 otherwise. The +2+2+2 is inside the last quantum integer, after taking the half.

Under all the preceding hypotheses, the statement asserts their conjunction:

  1. For every natural a,b,la,b,la,b,l, the closed rung satisfies
Dl;((slab;Kab;fabl)▹Vl);Cl=ι([a][b]) T(a,a,b,b;l,2).D_l;\bigl((s_l^{ab};K_{ab};f_{ab}^{l})\triangleright V_l\bigr);C_l =\iota([a][b])\,T(a,a,b,b;l,2).Dl​;((slab​;Kab​;fabl​)▹Vl​);Cl​=ι([a][b])T(a,a,b,b;l,2).

In particular the right-hand tetrahedron expands as Xabba;l;Habba;2RX_{abba;l};H^R_{abba;2}Xabba;l​;Habba;2R​. This is a claimed graphical closure identification, not an additional hypothesis; the separate seed only evaluates its already-defined right-hand side as a scalar. 2. For every natural a,b,la,b,la,b,l,

slab;Kab=ι(λab(l)) slab.s_l^{ab};K_{ab}=\iota(\lambda_{ab}(l))\,s_l^{ab}.slab​;Kab​=ι(λab​(l))slab​.
  1. For every natural a,b,c,d,la,b,c,d,la,b,c,d,l,
Xabcd;l;K^abcd=ι(λab(l)) Xabcd;l.X_{abcd;l};\widehat K_{abcd}=\iota(\lambda_{ab}(l))\,X_{abcd;l}.Xabcd;l​;Kabcd​=ι(λab​(l))Xabcd;l​.

There are no admissibility or nonzero-state hypotheses restricting these quantifiers. In particular the last two equations do not assert that the displayed maps are nonzero eigenvectors, nor that they form a basis. No recurrence or global tetrahedral evaluation is assumed.

Division and logical scope

All scalar divisions are total field divisions in rational-function fields: 0−1=00^{-1}=00−1=0, hence x/0=0x/0=0x/0=0. The formulas impose no implicit denominator-nonvanishing hypotheses. Negative factorials and inadmissible channels are deliberate zero extensions. These conventions do not define evaluation at a complex pole or assert a root-of-unity specialization. The three identities are the asserted theorem statement; no proof of them is supplied here.

Preamble
import Definitions.Def_QuantumSixJNetwork

open CategoryTheory MonoidalCategory
open scoped BigOperators QuantumSixJ

universe v u
Formal statement
namespace QuantumSixJ.Network

theorem rung_eigenvalue {C : Type u} [Category.{v} C] [Preadditive C] [Linear K C]
    [MonoidalCategory C] [MonoidalPreadditive C] [MonoidalLinear K C]
    (S : SuQ2Data C) (W : Vertex S) (hS : SuQ2Axioms S) (hW : VertexNormalized W) (hSeed : TetSeed W) :
    (∀ a b l : ℕ,
      S.cup l ≫ ((W.split l a b ≫ W.K2 a b ≫ W.fuse a b l) ▷ S.V l) ≫ S.cap l =
        ι (⟬(a : ℤ)⟭ * ⟬(b : ℤ)⟭) • W.tet a a b b l 2) ∧
    (∀ a b l : ℕ,
      W.split l a b ≫ W.K2 a b = ι (Λ a b l) • W.split l a b) ∧
    (∀ a b c d l : ℕ,
      W.X a b c d l ≫ W.Kop a b c d = ι (Λ a b l) • W.X a b c d l) := by sorry

end QuantumSixJ.Network
Source
I. Khavkine, Recurrence relation for the 6j-symbol of su_q(2) as a symmetric eigenvalue problem, International Journal of Geometric Methods in Modern Physics (2015), https://doi.org/10.1142/S0219887815501170; https://arxiv.org/abs/1009.2261. Eqs. (12)–(15), rung operator, triangle evaluation and eigenvalue.
Human review
  • Endorsed by Shuze Chen · Nov 19, 2026

    Confirmed by the moderator at approval.

  • Endorsed by igor.kh · Nov 19, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me