Prove2Me
Navigate
DiscoverCollectionsFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

§4.1, eq. (13), p. 34 — the local construction F(X, A)ᵤ = φ(xᵤ, X_{Nᵤ}) satisfies (10)

Proved
GeomDL.Sets.eq_13_equivariant

by mikedeng1 · Oct 21, 2026 · Mathlib 0df444a (Lean v4.33.1)

geometric-deep-learninggraph-neural-networksp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-bookp2o-v1permutation-equivariance

Let A∈Rn×n\mathbf A\in\mathbb R^{n\times n}A∈Rn×n be an adjacency matrix and X∈Rn×d\mathbf X\in\mathbb R^{n\times d}X∈Rn×d node features. For a node uuu let Nu={v:auv≠0 or avu≠0}\mathcal N_u=\{v:a_{uv}\ne0\text{ or }a_{vu}\ne0\}Nu​={v:auv​=0 or avu​=0} be its neighbourhood (11) and XNu={ ⁣{xv:v∈Nu} ⁣}\mathbf X_{\mathcal N_u}=\{\!\{\mathbf x_v:v\in\mathcal N_u\}\!\}XNu​​={{xv​:v∈Nu​}} the multiset of its neighbours' features (12). Let ϕ:Rd×{finite multisets of Rd}→Rd′\phi:\mathbb R^d\times\{\text{finite multisets of }\mathbb R^d\}\to\mathbb R^{d'}ϕ:Rd×{finite multisets of Rd}→Rd′ be any function, and let F(X,A)\mathbf F(\mathbf X,\mathbf A)F(X,A) be the matrix whose uuu-th row is ϕ(xu,XNu)\phi(\mathbf x_u,\mathbf X_{\mathcal N_u})ϕ(xu​,XNu​​), eq. (13). Then F\mathbf FF is permutation equivariant in the sense of (10):

F(PX,PAP⊤)=P F(X,A)for every permutation matrix P.\mathbf F(\mathbf P\mathbf X,\mathbf P\mathbf A\mathbf P^\top)=\mathbf P\,\mathbf F(\mathbf X,\mathbf A)\quad\text{for every permutation matrix }\mathbf P.F(PX,PAP⊤)=PF(X,A)for every permutation matrix P.

This is the book's general recipe for permutation-equivariant functions on graphs: a shared local function that is invariant to the order of the neighbours.

Formalization Note Since ϕ\phiϕ receives a multiset, its output does not depend on the ordering of the nodes in Nu\mathcal N_uNu​ — the book's hypothesis "ϕ\phiϕ is permutation invariant" holds by construction. Edges are the nonzero entries of A\mathbf AA, which for the 0/1 matrix (8) is the book's edge set.

Preamble
import Mathlib
import Definitions.Def_GeomDL_Sets_Defs
Formal statement
namespace GeomDL.Sets

/-- §4.1, eq. (13), p. 34: the local construction `F(X, A)_u = φ(x_u, X_{N_u})` with a shared
`φ` that receives the neighbourhood features as a multiset (hence is permutation invariant in
them) satisfies (10): `F(PX, PAPᵀ) = P F(X, A)` for every permutation matrix `P`. -/
theorem eq_13_equivariant {n d d' : ℕ}
    (φ : (Fin d → ℝ) → Multiset (Fin d → ℝ) → (Fin d' → ℝ)) :
    IsGraphPermEquivariant (localLayer φ : Matrix (Fin n) (Fin d) ℝ → Matrix (Fin n) (Fin n) ℝ →
      Matrix (Fin n) (Fin d') ℝ) := by sorry

end GeomDL.Sets
Source
Bronstein et al., Geometric Deep Learning (arXiv:2104.13478v2), §4.1, eqs. (11)–(13), p. 34
Human review
  • Endorsed by Shuze Chen · Oct 21, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 21, 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 paper
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me