Prove2Me
Navigate
DiscoverCollectionsFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

§4.3, p. 41 — group convolution against an invariant measure is equivariant

Proved
GeomDL.GroupConv.eq_14_equivariant

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

equivariancegeometric-deep-learninggroup-convolutionp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-bookp2o-v1

Let GGG act transitively and measurably on a domain Ω\OmegaΩ, and let μ\muμ be invariant under the action. For scalar signals x,θ:Ω→Rx,\theta:\Omega\to\mathbb Rx,θ:Ω→R, a group element hhh, and an output index g∈Gg\in Gg∈G, group convolution satisfies

((ρ(h)x)⋆θ)(g)=(x⋆θ)(h−1g)=(ρG(h)(x⋆θ))(g).((\rho(h)x)\star\theta)(g) =(x\star\theta)(h^{-1}g) =(\rho_G(h)(x\star\theta))(g).((ρ(h)x)⋆θ)(g)=(x⋆θ)(h−1g)=(ρG​(h)(x⋆θ))(g).

Thus the convolution maps signals on Ω\OmegaΩ to signals on GGG while respecting the two corresponding group actions.

Formalization Note In Lean the stated equality is between the first and last expressions; the middle expression is the definition of the regular action ρG\rho_GρG​. Transitivity records the section's homogeneous-domain assumption on p. 43. The book's invariant measure is explicit, as is measurability of the action. No integrability hypothesis is added; measure-preserving reindexing preserves the integral's value, including Lean's zero value for a nonintegrable integrand.

Preamble
import Mathlib
import Definitions.Def_GeomDL_GroupConv_Model
Formal statement
namespace GeomDL.GroupConv

universe u v

/-- §4.3, equation (14) and the following displayed identity, p. 41. -/
theorem eq_14_equivariant {G : Type u} {Ω : Type v} [Group G] [MulAction G Ω]
    [MeasurableSpace G] [MeasurableSpace Ω] [MeasurableSMul G Ω]
    (μ : MeasureTheory.Measure Ω) [MeasureTheory.SMulInvariantMeasure G Ω μ]
    (htrans : ∀ a b : Ω, ∃ k : G, k • a = b)
    (h : G) (x θ : Ω → ℝ) (g : G) :
    groupConvolution μ (signalAction h x) θ g =
      regularAction h (groupConvolution μ x θ) g := by sorry

end GeomDL.GroupConv
Source
Bronstein et al., Geometric Deep Learning (arXiv:2104.13478v2), §4.3, equation (14) and displayed equivariance identity, p. 41
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