Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§7.2: after the first iteration ∑G(i,j)=g(yik)j=0\sum_{\mathcal G(i,j)=g}(y_i^k)_j=0∑G(i,j)=g​(yik​)j​=0 and zgk+1=(1/kg)∑G(i,j)=g(xik+1)jz_g^{k+1}=(1/k_g)\sum_{\mathcal G(i,j)=g}(x_i^{k+1})_jzgk+1​=(1/kg​)∑G(i,j)=g​(xik+1​)j​

Proved
BoydADMM.Consensus.general_dual_sums_zero

by mikedeng1 · Oct 6, 2026 · Mathlib 0df444a (Lean v4.33.1)

admmconsensusdual-variablesp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-bookp2o-v1

Let xik,zk,yikx_i^k, z^k, y_i^kxik​,zk,yik​ be any run of general form consensus ADMM (§7.2) with ρ>0\rho>0ρ>0, and assume kg≥1k_g\ge1kg​≥1 for every global index ggg. Then, for every k≥1k\ge1k≥1 and every ggg:

  1. the dual entries attached to ggg sum to zero:
∑G(i,j)=g(yik)j=0;\sum_{\mathcal G(i,j)=g}(y_i^k)_j=0;G(i,j)=g∑​(yik​)j​=0;
  1. the zzz-update is plain local averaging:
zgk+1=1kg∑G(i,j)=g(xik+1)j.z_g^{k+1}=\frac1{k_g}\sum_{\mathcal G(i,j)=g}(x_i^{k+1})_j.zgk+1​=kg​1​G(i,j)=g∑​(xik+1​)j​.

Formalization Note "After the first iteration" is read as k≥1k\ge1k≥1 (the run starts at an arbitrary y0y^0y0). The hypothesis kg≥1k_g\ge1kg​≥1 is implicit in the book. The run's zzz-update is the argmin of the page's zzz-objective, so item 2 is derived, not assumed.

Preamble
import Mathlib
import Definitions.Def_BoydADMM_Consensus_Model

open scoped BigOperators InnerProductSpace
Formal statement
namespace BoydADMM.Consensus

/-- §7.2, p. 55: along every run of general form consensus ADMM with `ρ > 0` and every
`k_g ≥ 1`, after the first iteration (`k ≥ 1`) the dual entries attached to each global index
sum to zero, `∑_{G(i,j)=g} (y_i^k)_j = 0`, and the `z`-update is plain local averaging,
`z_g^{k+1} = (1/k_g) ∑_{G(i,j)=g} (x_i^{k+1})_j`. -/
theorem general_dual_sums_zero {n N : ℕ} {nl : Fin N → ℕ}
    (G : (i : Fin N) → Fin (nl i) → Fin n) (hG : ∀ g, 1 ≤ kg G g) (ρ : ℝ) (hρ : 0 < ρ)
    (Cf : (i : Fin N) → Set (EuclideanSpace ℝ (Fin (nl i))))
    (f : (i : Fin N) → EuclideanSpace ℝ (Fin (nl i)) → ℝ)
    (x : ℕ → (i : Fin N) → EuclideanSpace ℝ (Fin (nl i))) (z : ℕ → EuclideanSpace ℝ (Fin n))
    (y : ℕ → (i : Fin N) → EuclideanSpace ℝ (Fin (nl i)))
    (hrun : IsGeneralConsensusADMMRun G Cf f ρ x z y) :
    (∀ k, 1 ≤ k → ∀ g, ∑ p ∈ entriesOf G g, y k p.1 p.2 = 0) ∧
    (∀ k, 1 ≤ k → ∀ g,
      z (k + 1) g = (1 / (kg G g : ℝ)) * ∑ p ∈ entriesOf G g, x (k + 1) p.1 p.2) := by sorry

end BoydADMM.Consensus
Source
Boyd, Parikh, Chu, Peleato, Eckstein, Distributed Optimization and Statistical Learning via the Alternating Direction Method of Multipliers, Found. Trends Mach. Learn. 3(1) (2011), p. 55, §7.2 (∑_{𝒢(i,j)=g} (y_i^k)_j = 0 after the first iteration; simplified z-update)
Human review
  • Endorsed by Shuze Chen · Oct 6, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 6, 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