Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 3.1, p. 14 — on the instance (3.10), the direct extension of ADMM diverges for every β > 0 from an open half of a 3-dimensional subspace of starting points

Proved
ExtADMM.Diverge.theorem_3_1

by mikedeng1 · Oct 7, 2026 · Mathlib 0df444a (Lean v4.33.1)

admmcounterexampledivergencep2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1spectral-radius

Consider the three-block problem (1.1) given by the linear system

A1x1+A2x2+A3x3=0,A=(A1,A2,A3)=(111112122),xi∈R,A_1x_1+A_2x_2+A_3x_3=0,\qquad A=(A_1,A_2,A_3)=\begin{pmatrix}1&1&1\\1&1&2\\1&2&2\end{pmatrix},\quad x_i\in\mathbb R,A1​x1​+A2​x2​+A3​x3​=0,A=(A1​,A2​,A3​)=​111​112​122​​,xi​∈R,

with null objective, b=0b=0b=0 and Xi=R\mathcal X_i=\mathbb RXi​=R (example (3.10)). Then:

  1. this instance satisfies the standing assumptions of (1.1) (closed convex sets, convex objectives, nonempty solution set);
  2. there are a three-dimensional real subspace SSS of R5\mathbb R^5R5 and a nonzero linear functional φ\varphiφ on SSS such that, for every penalty parameter β>0\beta>0β>0 and every starting point (x20,x30,λ0)∈R×R×R3(x_2^0,x_3^0,\lambda^0)\in\mathbb R\times\mathbb R\times\mathbb R^3(x20​,x30​,λ0)∈R×R×R3 with
z0=(x20, x30, λ0/β)∈Sandφ(z0)>0,z^0=\bigl(x_2^0,\ x_3^0,\ \lambda^0/\beta\bigr)\in S\quad\text{and}\quad\varphi(z^0)>0,z0=(x20​, x30​, λ0/β)∈Sandφ(z0)>0,

the direct extension of ADMM (1.5) with penalty β\betaβ has a run starting from (x20,x30,λ0)(x_2^0,x_3^0,\lambda^0)(x20​,x30​,λ0), and no run from that starting point converges: the sequence (x1k,x2k,x3k,λk)(x_1^k,x_2^k,x_3^k,\lambda^k)(x1k​,x2k​,x3k​,λk) has no limit.

This is the paper's main theorem. It answers negatively the question whether the direct extension of the two-block ADMM to three blocks converges for convex problems: even on a nonsingular linear system with zero objective, for every β>0\beta>0β>0 and from an open half of a three-dimensional subspace of starting points, the iteration diverges.

Formalization Note. "There is an example" is witnessed by the paper's own instance (3.10). The paper's "continuously dense half space of dimension 3" is {z∈S:φ(z)>0}\{z\in S:\varphi(z)>0\}{z∈S:φ(z)>0} in the coordinates (x20,x30,μ0)(x_2^0,x_3^0,\mu^0)(x20​,x30​,μ0) with μ0=λ0/β\mu^0=\lambda^0/\betaμ0=λ0/β of (3.12)–(3.13); in these coordinates the set does not depend on β\betaβ, so SSS and φ\varphiφ are chosen before β\betaβ. "Divergent" is read as "does not converge" (in the product topology of R×R×R×R3\mathbb R\times\mathbb R\times\mathbb R\times\mathbb R^3R×R×R×R3), and the existence of a run is asserted so that the claim is not vacuous. x10x_1^0x10​ is not used by (1.5).

Preamble
import Mathlib
import Definitions.Def_ExtADMM_Diverge_Setting

open Matrix Filter Topology
Formal statement
namespace ExtADMM.Diverge

/-- Theorem 3.1, p. 14. The instance (3.10) of problem (1.1) satisfies the standing
assumptions, and there are a 3-dimensional subspace `S` of the space of starting points
`(x₂⁰, x₃⁰, λ⁰/β)` and a nonzero linear functional `φ` on `S` such that, for every penalty
`β > 0` and every starting point `(x₂⁰, x₃⁰, λ⁰)` with `(x₂⁰, x₃⁰, λ⁰/β)` in the open half
`{z ∈ S | φ z > 0}`, the direct extension of ADMM (1.5) has a run from that point, and no
run from that point converges. -/
theorem theorem_3_1 :
    example310.Standing ∧
    ∃ S : Submodule ℝ (Fin 5 → ℝ), Module.finrank ℝ S = 3 ∧ ∃ φ : S →ₗ[ℝ] ℝ, φ ≠ 0 ∧
      ∀ β : ℝ, 0 < β → ∀ (x20 x30 : Fin 1 → ℝ) (lam0 : Fin 3 → ℝ)
        (hz : stateVec β x20 x30 lam0 ∈ S), 0 < φ ⟨_, hz⟩ →
          (∃ (x1 x2 x3 : ℕ → Fin 1 → ℝ) (lam : ℕ → Fin 3 → ℝ),
              example310.IsRun15 β x1 x2 x3 lam ∧ x2 0 = x20 ∧ x3 0 = x30 ∧ lam 0 = lam0) ∧
          ∀ (x1 x2 x3 : ℕ → Fin 1 → ℝ) (lam : ℕ → Fin 3 → ℝ),
            example310.IsRun15 β x1 x2 x3 lam → x2 0 = x20 → x3 0 = x30 → lam 0 = lam0 →
              ¬ ∃ w, Tendsto (fun k => (x1 k, x2 k, x3 k, lam k)) atTop (𝓝 w) := by sorry

end ExtADMM.Diverge
Source
Chen, He, Ye & Yuan, The direct extension of ADMM for multi-block convex minimization problems is not necessarily convergent, Math. Program., DOI 10.1007/s10107-014-0826-5 (authors' version of January 22, 2014), p. 14, Theorem 3.1
Human review
  • Endorsed by Shuze Chen · Oct 8, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 8, 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