Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

S2.6, p. 27 — EvNorm (S39): h̄ is invariant and ĥ is equivariant under the unitary representations U^L

Proved
OrbNetEqui.Equivariance.evNorm_equivariant

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

equivariancemachine-learningnormalizationp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1

Let (UL)L(U^L)_{L}(UL)L​ be unitary representations of a group KKK on VL=Cdim⁡L\mathbb V^L=\mathbb C^{\dim L}VL=CdimL, and let xxx be a feature, i.e. a vector xL,c∈VLx_{L,c}\in\mathbb V^LxL,c​∈VL for each label LLL and channel ccc. EvNorm (S39) with statistics μL,c,σL,c\mu_{L,c},\sigma_{L,c}μL,c​,σL,c​, positive scales βL,c>0\beta_{L,c}>0βL,c​>0 and stability factor ϵ>0\epsilon>0ϵ>0 is

xˉL,c=∥xL,c∥−μL,cσL,c,x^L,c,M=xL,c,M∥xL,c∥+1/βL,c+ϵ,\bar x_{L,c}=\frac{\|x_{L,c}\|-\mu_{L,c}}{\sigma_{L,c}},\qquad \hat x_{L,c,M}=\frac{x_{L,c,M}}{\|x_{L,c}\|+1/\beta_{L,c}+\epsilon},xˉL,c​=σL,c​∥xL,c​∥−μL,c​​,x^L,c,M​=∥xL,c​∥+1/βL,c​+ϵxL,c,M​​,

with the regularized norm ∥x∥=∑M∣xM∣2+ϵ2−ϵ\|x\|=\sqrt{\sum_M|x_M|^2+\epsilon^2}-\epsilon∥x∥=∑M​∣xM​∣2+ϵ2​−ϵ of (23).

Then for every k∈Kk\in Kk∈K, writing (k⋅x)L,c=UkL xL,c(k\cdot x)_{L,c}=U^L_k\,x_{L,c}(k⋅x)L,c​=UkL​xL,c​,

k⋅x‾=xˉandk⋅x^=k⋅x^.\overline{k\cdot x}=\bar x\qquad\text{and}\qquad\widehat{k\cdot x}=k\cdot\hat x.k⋅x=xˉandk⋅x=k⋅x^.

The invariant part feeds the multilayer perceptrons of the point-wise interaction block, and the equivariant part carries the directions.

Formalization Note. Scalars are complex, so the norm uses ∣xM∣2|x_M|^2∣xM​∣2. The statistics μ,σ\mu,\sigmaμ,σ are fixed parameters, so the statement holds for every value of them (batch or layer statistics alike). The page's positivity of β\betaβ and ϵ\epsilonϵ is kept as hypotheses. No subgroup restriction is needed, because every UkLU^L_kUkL​ is unitary.

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

open Matrix

theorem evNorm_equivariant {K : Type*} [Group K] {Λ : Type*} {dim : Λ → ℕ} {nmax : ℕ}
    (U : (L : Λ) → K →* Matrix.unitaryGroup (Fin (dim L)) ℂ)
    (μ σ β : Λ → Fin nmax → ℝ) (ε : ℝ) (hβ : ∀ L c, 0 < β L c) (hε : 0 < ε)
    (k : K) (x : Feature Λ dim nmax) :
    evNormBar μ σ ε (featAct U k x) = evNormBar μ σ ε x ∧
      evNormHat β ε (featAct U k x) = featAct U k (evNormHat β ε x) := by sorry

end OrbNetEqui.Equivariance
Source
Qiao et al., Informing Geometric Deep Learning with Electronic Interactions to Accelerate Quantum Chemistry, arXiv:2105.14655v4, p. 27, S2.6, "Proof of G-equivariance for EvNorm (S39)"; eqs. (22)–(23) p. 9, (S39) p. 26
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