Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

S2.6, p. 27 — message passing (S37)–(S38) is equivariant: m̃_{u₁} ↦ π¹(g_{u₁}) m̃_{u₁}, and ρ†_{u₁}(m̃) ↦ U(g_{u₁}) ρ†_{u₁}(m̃)

Proved
OrbNetEqui.Equivariance.messagePassing_equivariant

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

equivariancemachine-learningmessage-passingp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1

Assume the setting of the block-convolution milestone: subgroups GuG_uGu​ of a group KKK, unitary representations πu1\pi^1_uπu1​ and ULU^LUL, and matching matrices satisfying (S35) Duπu2(k)=πu1(k)DuD_u\pi^2_u(k)=\pi^1_u(k)D_uDu​πu2​(k)=πu1​(k)Du​ for k∈Guk\in G_uk∈Gu​. Fix scalar attention weights αu⃗j\alpha^j_{\vec u}αuj​; using the same α\alphaα before and after the transformation expresses the paper's invariance condition g⋅α=αg\cdot\alpha=\alphag⋅α=α.

For g=(gu)ug=(g_u)_ug=(gu​)u​ with gu∈Gug_u\in G_ugu​∈Gu​, any tensor TTT, features hhh and atom u1u_1u1​, the aggregated message (S37)

m~u1=∑u2,…,uN⨁i,j(mu⃗)i αu⃗j\tilde{\mathbf m}_{u_1}=\sum_{u_2,\dots,u_N}\bigoplus_{i,j}(\mathbf m_{\vec u})^i\,\alpha^j_{\vec u}m~u1​​=u2​,…,uN​∑​i,j⨁​(mu​)iαuj​

satisfies:

  1. m~u1(g⋅T,g⋅h)(i,j)=πu11(gu1) m~u1(T,h)(i,j)\tilde{\mathbf m}_{u_1}(g\cdot T,g\cdot h)^{(i,j)}=\pi^1_{u_1}(g_{u_1})\,\tilde{\mathbf m}_{u_1}(T,h)^{(i,j)}m~u1​​(g⋅T,g⋅h)(i,j)=πu1​1​(gu1​​)m~u1​​(T,h)(i,j) for every (i,j)(i,j)(i,j);
  2. the reverse matching layer (S45) turns it into an equivariant feature: ρu1†(m~u1(g⋅T,g⋅h))=gu1⋅ρu1†(m~u1(T,h))\rho^\dagger_{u_1}\big(\tilde{\mathbf m}_{u_1}(g\cdot T,g\cdot h)\big)=g_{u_1}\cdot\rho^\dagger_{u_1}\big(\tilde{\mathbf m}_{u_1}(T,h)\big)ρu1​†​(m~u1​​(g⋅T,g⋅h))=gu1​​⋅ρu1​†​(m~u1​​(T,h)).

Together these give the equivariance of the second argument of the update (S38).

Formalization Note. The page displays only the first conclusion. The second one, about the reverse matching layer (S45), is the step that (S38) also needs, and it is added to complete the page's claim for "(S37)-(S38)". The reverse matching weights have the main-text shape (28), WL†∈Cnmax⁡×KLIJW^\dagger_L\in\mathbb C^{n_{\max}\times K_L I J}WL†​∈Cnmax​×KL​IJ, and may depend on the atom.

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

open Matrix

theorem messagePassing_equivariant {K : Type*} [Group K] {d n : ℕ} {ι : Fin d → Type*}
    [∀ u, Fintype (ι u)] [∀ u, DecidableEq (ι u)] {Λ : Type*} [Fintype Λ] [DecidableEq Λ]
    {dim : Λ → ℕ} {nmax I J : ℕ} {Kmul : Fin d → Λ → ℕ}
    (S : Fin d → Subgroup K)
    (π₁ : (u : Fin d) → K →* Matrix.unitaryGroup (ι u) ℂ)
    (U : (L : Λ) → K →* Matrix.unitaryGroup (Fin (dim L)) ℂ)
    (D : (u : Fin d) → Matrix (ι u) (MIdx dim Kmul u) ℂ)
    (W : (u : Fin d) → Fin I → (L : Λ) → Matrix (Fin (Kmul u L)) (Fin nmax) ℂ)
    (Wd : (u : Fin d) → (L : Λ) → Matrix (Fin nmax) (Fin (Kmul u L) × Fin I × Fin J) ℂ)
    (hD : ∀ u, ∀ k ∈ S u, D u * π₂mat U Kmul u k = (π₁ u k : Matrix (ι u) (ι u) ℂ) * D u)
    (g : Fin d → K) (hg : ∀ u, g u ∈ S u)
    (T : (Fin (n + 1) → Orb ι) → ℂ) (h : Fin d → Feature Λ dim nmax)
    (α : (Fin (n + 1) → Fin d) → Fin J → ℂ) (u : Fin d) :
    aggregate D W (tensorAct π₁ g T) (fun w => featAct U (g w) (h w)) α u =
        (fun ij => (π₁ u (g u) : Matrix (ι u) (ι u) ℂ) *ᵥ aggregate D W T h α u ij) ∧
      revMatching D Wd u (aggregate D W (tensorAct π₁ g T) (fun w => featAct U (g w) (h w)) α u) =
        featAct U (g u) (revMatching D Wd u (aggregate D W T h α u)) := 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 the message passing block (S37)-(S38)"; eqs. (S37), (S38), (S45)
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