Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Third-order equality strata and 203 linear parameters

Open
QAlgorithms.Papers.Independent.Intakenext100.P94.third_order_parameters

by Gohan · Oct 31, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

Let n≥6n\ge6n≥6 and order the seven shapes as (n),(n−1,1),(n−2,2),(n−2,1,1),(n−3,3),(n−3,2,1),(n−3,1,1,1)(n),(n-1,1),(n-2,2),(n-2,1,1),(n-3,3),(n-3,2,1),(n-3,1,1,1)(n),(n−1,1),(n−2,2),(n−2,1,1),(n−3,3),(n−3,2,1),(n−3,1,1,1). In the full real order-three tensor representation, split the coordinates into pairwise distinct indices, three exactly-two-equal patterns, and all-equal indices. Their respective multiplicity vectors are (1,3,3,3,1,2,1)(1,3,3,3,1,2,1)(1,3,3,3,1,2,1), (1,2,1,1,0,0,0)(1,2,1,1,0,0,0)(1,2,1,1,0,0,0) for each of the three middle strata, and (1,1,0,0,0,0,0)(1,1,0,0,0,0,0)(1,1,0,0,0,0,0). Consequently the full type is (5,10,6,6,1,2,1)(5,10,6,6,1,2,1)(5,10,6,6,1,2,1) and

dim⁡REnd⁡Sn(Rn×n×n)=52+102+62+62+12+22+12=203.\dim_{\mathbb R}\operatorname{End}_{S_n}(\mathbb R^{n\times n\times n})=5^2+10^2+6^2+6^2+1^2+2^2+1^2=203.dimR​EndSn​​(Rn×n×n)=52+102+62+62+12+22+12=203.

Each type is an actual equivariant linear isomorphism to copies of concrete real Specht modules. The count concerns the linear operation before any nonlinearity.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P94
Formal statement
noncomputable section

namespace QAlgorithms.Papers.Independent.Intakenext100.P94

open scoped BigOperators

attribute [local instance] Classical.propDecidable

/-- Appendix C §10.1, source PDF p.15: all five equality strata, full type and
203-dimensional space of linear layers. Stable range n≥6 is explicit; the
printed count fails for small n. The dimensions count maps, never scalar evaluations. -/
theorem third_order_parameters (n : ℕ) (hn : 6 ≤ n)
    (lambda : Fin 7 → QAlgorithms.Schur.Partition n n)
    (hshape : ∀ a, (lambda a).val.rowLens = thirdShapes n a) :
    HasTensorType (DistinctSpace n 3) lambda ![1,3,3,3,1,2,1] ∧
    (∀ a : Fin 3, HasTensorType (supportedSpace (pairPattern a)) lambda ![1,2,1,1,0,0,0]) ∧
    HasTensorType (supportedSpace (fun t : Fin 3 → Fin n => ∀ a b, t a = t b))
      lambda ![1,1,0,0,0,0,0] ∧
    HasTensorType (⊤ : Submodule ℝ (Tensor n 3)) lambda ![5,10,6,6,1,2,1] ∧
    Module.finrank ℝ (equivariantLinearMaps (permutationTensorAction (n := n) (k := 3))
      (permutationTensorAction (n := n) (k := 3))) = 203 := by sorry

end QAlgorithms.Papers.Independent.Intakenext100.P94

end
Source
The general theory of permutation equivarant neural networks and higher order graph variational encoders, arXiv:2004.03990v1, PDF p. 15, §10.1 (Appendix C): third order decomposition and 203 parameters
Read-back

What the Lean code literally says, in plain math · gpt-6.1-sol

Write [m]={0,…,m−1}[m]=\{0,\ldots,m-1\}[m]={0,…,m−1}. For every natural number n≥6n\ge6n≥6 and every family λ0,…,λ6\lambda_0,\ldots,\lambda_6λ0​,…,λ6​ of Young diagrams with nnn cells and first-column length at most nnn, assume their row-length lists are respectively [n][n][n], [n−1,1][n-1,1][n−1,1], [n−2,2][n-2,2][n−2,2], [n−2,1,1][n-2,1,1][n−2,1,1], [n−3,3][n-3,3][n−3,3], [n−3,2,1][n-3,2,1][n−3,2,1], and [n−3,1,1,1][n-3,1,1,1][n−3,1,1,1]. Let TTT be the real vector space of all functions on triples t:[3]→[n]t:[3]\to[n]t:[3]→[n], with permutation action (AsF)(t)=F(s−1∘t)(A_sF)(t)=F(s^{-1}\circ t)(As​F)(t)=F(s−1∘t). For each λ\lambdaλ, use the fixed numbering of its cells by [n][n][n], let Rλ,Cλ⊆SnR_\lambda,C_\lambda\subseteq S_nRλ​,Cλ​⊆Sn​ consist of permutations preserving every numbered cell’s row and column, respectively, and define Vλ⊆RSnV_\lambda\subseteq\mathbb R^{S_n}Vλ​⊆RSn​ as the range of the matrix

Yλ(g,h)=∑a∈Rλ∑b∈Cλ{sgn⁡(b),g=hab,0,g≠hab.Y_\lambda(g,h)=\sum_{a\in R_\lambda}\sum_{b\in C_\lambda} \begin{cases}\operatorname{sgn}(b),&g=hab,\\0,&g\ne hab.\end{cases}Yλ​(g,h)=a∈Rλ​∑​b∈Cλ​∑​{sgn(b),0,​g=hab,g=hab.​

For any subspace S⊆TS\subseteq TS⊆T and multiplicity vector τ\tauτ, a decomposition assertion here means that there exists a real-linear isomorphism U:S→∏b=06VλbτbU:S\to\prod_{b=0}^{6}V_{\lambda_b}^{\tau_b}U:S→∏b=06​Vλb​τb​​ such that, for every s∈Sns\in S_ns∈Sn​ and F∈SF\in SF∈S, there exists G∈SG\in SG∈S with G(t)=F(s−1∘t)G(t)=F(s^{-1}\circ t)G(t)=F(s−1∘t) for every triple ttt, and (UG)b,c(g)=(UF)b,c(s−1g)(UG)_{b,c}(g)=(UF)_{b,c}(s^{-1}g)(UG)b,c​(g)=(UF)b,c​(s−1g) for every b∈[7]b\in[7]b∈[7], c∈[τb]c\in[\tau_b]c∈[τb​], and g∈Sng\in S_ng∈Sn​. The theorem asserts all of the following: this decomposition holds for functions vanishing on all triples that are not injective, with multiplicities (1,3,3,3,1,2,1)(1,3,3,3,1,2,1)(1,3,3,3,1,2,1); for every a∈[3]a\in[3]a∈[3], it holds for functions vanishing outside the triples satisfying ta+1=ta+2t_{a+1}=t_{a+2}ta+1​=ta+2​ and ta≠ta+1t_a\ne t_{a+1}ta​=ta+1​, with indices modulo 333, with multiplicities (1,2,1,1,0,0,0)(1,2,1,1,0,0,0)(1,2,1,1,0,0,0); it holds for functions vanishing outside triples whose three entries are all equal, with multiplicities (1,1,0,0,0,0,0)(1,1,0,0,0,0,0)(1,1,0,0,0,0,0); and it holds for all of TTT, with multiplicities (5,10,6,6,1,2,1)(5,10,6,6,1,2,1)(5,10,6,6,1,2,1). A zero multiplicity contributes no copies and no coordinate conditions. Finally, the real vector space of real-linear maps L:T→TL:T\to TL:T→T satisfying L(AsF)=As(LF)L(A_sF)=A_s(LF)L(As​F)=As​(LF) for every s∈Sns\in S_ns∈Sn​ and F∈TF\in TF∈T has dimension exactly 203203203.

Human review
  • Endorsed by Shuze Chen · Oct 31, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Oct 31, 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