Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Corollary S3, p. 24 — the diagonal reduction ψ(T)=Σ T(v⃗)Q^{v⃗} is injective, Σ_L n_L ≤ dim(V_u)^N, and ψ(g·T)_{ν,L} = U^L_g ψ(T)_{ν,L}

Proved
OrbNetEqui.Equivariance.corollary_S3

by mikedeng1 · Oct 22, 2026 · Mathlib 0df444a (Lean v4.33.1)

equivariancemachine-learningp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1representation-theory

In the setting of Corollary S2 (KKK a group, (UL)L(U^L)_L(UL)L​ a complete catalogue of irreducible unitary representations, Vu≃⨁aVlab(a)V_u\simeq\bigoplus_a\mathbb V^{\mathrm{lab}(a)}Vu​≃⨁a​Vlab(a) with the block-diagonal action π2\pi^2π2, and KKK acting on Vu⊗NV_u^{\otimes N}Vu⊗N​ diagonally), there exist numbers nL∈Nn_L\in\mathbb NnL​∈N, nonzero for finitely many LLL, and tensor-independent coefficients Qν,L,Mv⃗Q^{\vec v}_{\nu,L,M}Qν,L,Mv​ such that the linear map

(ψ(Tu))ν,L,M:=∑v⃗Tu(v1,…,vN) Qν,L,Mv⃗,ν∈{1,…,nL}(S31)\big(\psi(\mathbf T_u)\big)_{\nu,L,M}:=\sum_{\vec v}T_u(v_1,\dots,v_N)\,Q^{\vec v}_{\nu,L,M},\qquad \nu\in\{1,\dots,n_L\}\qquad\text{(S31)}(ψ(Tu​))ν,L,M​:=v∑​Tu​(v1​,…,vN​)Qν,L,Mv​,ν∈{1,…,nL​}(S31)

satisfies:

  1. ψ\psiψ is injective;
  2. ∑LnL≤dim⁡(Vu)N\sum_L n_L\le\dim(V_u)^N∑L​nL​≤dim(Vu​)N;
  3. for every k∈Kk\in Kk∈K, ψ(k⋅Tu)ν,L=UkL ψ(Tu)ν,L\psi(k\cdot\mathbf T_u)_{\nu,L}=U^L_k\,\psi(\mathbf T_u)_{\nu,L}ψ(k⋅Tu​)ν,L​=UkL​ψ(Tu​)ν,L​ (S32).

This is the equivariance of UNiTE's diagonal reduction layer, which the mission's goal theorem takes as a hypothesis on ψ\psiψ.

Formalization Note. U(n)\mathrm U(n)U(n) and GuG_uGu​ are replaced by an arbitrary group KKK with a complete catalogue, as in Theorem S1; equivariance is claimed for all of KKK. The corollary is stated, as in the paper, in the Definition S2 basis of VuV_uVu​; the goal theorem works in a general basis {eu;v}\{e_{u;v}\}{eu;v​} with an arbitrary unitary representation πu1\pi^1_uπu1​ and assumes (S32) for its ψu\psi_uψu​.

Preamble
import Mathlib
import Definitions.Def_OrbNetEqui_Equivariance_Basic
Formal statement
namespace OrbNetEqui.Equivariance

open Matrix

theorem corollary_S3 {K : Type*} [Group K] {Λ : Type*} {dim : Λ → ℕ}
    (U : (L : Λ) → K →* Matrix.unitaryGroup (Fin (dim L)) ℂ)
    (hirr : ∀ L, IsIrreducibleUnitary (U L)) (hcat : CatalogComplete U)
    {r : ℕ} (lab : Fin r → Λ) (N : ℕ) :
    ∃ (μ : Λ → ℕ) (s : Finset Λ), (∀ L, L ∉ s → μ L = 0) ∧
      ∃ Q : (L : Λ) → Fin (μ L) → Fin (dim L) → (Fin N → Σ a : Fin r, Fin (dim (lab a))) → ℂ,
        Function.Injective (coeffMap Q) ∧
        ∑ L ∈ s, μ L ≤ Fintype.card (Σ a : Fin r, Fin (dim (lab a))) ^ N ∧
        ∀ (k : K) (T : (Fin N → Σ a : Fin r, Fin (dim (lab a))) → ℂ) (L : Λ) (ν : Fin (μ L)),
          coeffMap Q (diagAct (labelRep U lab k) T) L ν =
            (U L k : Matrix (Fin (dim L)) (Fin (dim L)) ℂ) *ᵥ coeffMap Q T L ν := by sorry

end OrbNetEqui.Equivariance
Source
Qiao et al., Informing Geometric Deep Learning with Electronic Interactions to Accelerate Quantum Chemistry, arXiv:2105.14655v4, p. 24, Corollary S3, eqs. (S31)–(S32)
Human review
  • Endorsed by Shuze Chen · Oct 22, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 22, 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