Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Appendix G: tied QGCNN generators preserve graph automorphisms

Proved
QAlgorithms.Papers.Independent.Advance50.P50.qgcnn_automorphism_invariance

by Gohan · Oct 29, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

Let GGG be a fixed simple graph, ppp an automorphism of its vertices, and PPP its physical site-permutation matrix. Fix finite families of Hermitian one-qubit operators Or,RsO_r,R_sOr​,Rs​ and real parameters Wr,BsW_r,B_sWr​,Bs​, tied across all vertices and edges. Count each unordered edge once and define

H=∑{i,j}∈E(G)∑rWrOr(i)Or(j)+∑v∈V(G)∑sBsRs(v).H=\sum_{\{i,j\}\in E(G)}\sum_rW_r O_r^{(i)}O_r^{(j)}+\sum_{v\in V(G)}\sum_sB_s R_s^{(v)}.H={i,j}∈E(G)∑​r∑​Wr​Or(i)​Or(j)​+v∈V(G)∑​s∑​Bs​Rs(v)​.

Then

PHP†=H.PHP^\dagger=H.PHP†=H.

This is the automorphism invariance of the Appendix G generator, adapted there from Ref.[32]. It supplies no full-permutation invariance for a generic fixed graph.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Advance50_P50
Formal statement
noncomputable section

namespace QAlgorithms.Papers.Independent.Advance50.P50

open MeasureTheory ProbabilityTheory

open scoped ComplexConjugate ENNReal

-- Exact accepted primitives; no paper-local redeclarations.

open QAlgorithms.Papers.Independent.Information100.P63 (Mat tensorPower)

open QAlgorithms.Papers.Independent.Gap48.P20 (Copies)

open QAlgorithms.Papers.Independent.Gap48.P38 (subsystemReduction)

open QAlgorithms.Papers.Independent.Information100.P32 (purity)

open QAlgorithms.Papers.Independent.Information100.P08 (IsPure)

-- Matrices carry the associative commutator bracket X*Y-Y*X.

attribute [local instance 100] LieRing.ofAssociativeRing

/-- PDF p.29, Appendix G: tied generators commute with every graph
AUTOMORPHISM. This does not assert full-Sn equivariance for a fixed graph. -/
theorem qgcnn_automorphism_invariance {n a b : ℕ} (G : SimpleGraph (Fin n))
    (p : Equiv.Perm (Fin n)) (hp : ∀ i j, G.Adj (p i) (p j) ↔ G.Adj i j)
    (W : Fin a → ℝ) (B : Fin b → ℝ) (O : Fin a → Mat Bool) (R : Fin b → Mat Bool)
    (hO : ∀ r, (O r).IsHermitian) (hR : ∀ r, (R r).IsHermitian) :
    conjSuperop (sitePermutation p) (qgcnnGenerator G W B O R) = qgcnnGenerator G W B O R := by sorry

end QAlgorithms.Papers.Independent.Advance50.P50

end
Source
Group-Invariant Quantum Machine Learning, arXiv:2205.02261v2, PDF p. 29, Appendix G: tied QGCNN generators preserve graph automorphisms
Read-back

What the Lean code literally says, in plain math · gpt-6.1-sol

For arbitrary natural numbers n,a,bn,a,bn,a,b, let GGG be a simple graph on {0,…,n−1}\{0,\ldots,n-1\}{0,…,n−1}, and let ppp be a permutation of these vertices satisfying G(p(i),p(j))↔G(i,j)G(p(i),p(j))\leftrightarrow G(i,j)G(p(i),p(j))↔G(i,j) for every i,ji,ji,j. For arbitrary real coefficient families WrW_rWr​ and BrB_rBr​, and Hermitian complex 2×22\times22×2 matrix families OrO_rOr​ for r<ar<ar<a and RrR_rRr​ for r<br<br<b, define Mx,y(i)=Mxi,yi∏j≠i1{xj=yj}M^{(i)}_{x,y}=M_{x_i,y_i}\prod_{j\ne i}\mathbf1_{\{x_j=y_j\}}Mx,y(i)​=Mxi​,yi​​∏j=i​1{xj​=yj​}​ on nnn-bit strings and put

K=∑i<jG(i,j)∑r<aWr Or(i)Or(j)+∑v<n∑r<bBr Rr(v).K=\sum_{\substack{i<j\\G(i,j)}}\sum_{r<a}W_r\,O_r^{(i)}O_r^{(j)} +\sum_{v<n}\sum_{r<b}B_r\,R_r^{(v)}.K=i<jG(i,j)​∑​r<a∑​Wr​Or(i)​Or(j)​+v<n∑​r<b∑​Br​Rr(v)​.

Let PpP_pPp​ be the matrix with entry 111 at (y,x)(y,x)(y,x) exactly when y=x∘p−1y=x\circ p^{-1}y=x∘p−1, and 000 otherwise. The conclusion is

PpKPp†=K.P_p K P_p^\dagger=K.Pp​KPp†​=K.

All zero sizes are included; for n=0n=0n=0, K=0K=0K=0, and the graph and matrix-family hypotheses with empty index sets are vacuous.

Human review
  • Endorsed by Shuze Chen · Oct 29, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Oct 29, 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