Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§8.3.1, p. 69 (corrected) — the lasso block update is 0 iff ‖Aᵢᵀv‖_∞ ≤ λ/ρ

Proved
BoydADMM.ModelFit.lasso_block_zero_test_corrected

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

admmlassooptimality-conditionp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-bookp2o-v1

Let Ai∈Rm×niA_i\in\mathbb R^{m\times n_i}Ai​∈Rm×ni​, v∈Rmv\in\mathbb R^mv∈Rm, ρ>0\rho>0ρ>0 and λ>0\lambda>0λ>0. In the feature-split lasso the xix_ixi​-update minimizes ρ2∥Aix−v∥22+λ∥x∥1\frac{\rho}{2}\|A_ix-v\|_2^2+\lambda\|x\|_12ρ​∥Ai​x−v∥22​+λ∥x∥1​ with v=Aixik+zˉk−Ax‾k−ukv=A_ix_i^k+\bar z^k-\overline{Ax}^k-u^kv=Ai​xik​+zˉk−Axk−uk. Then x=0x=0x=0 is a minimizer of

ρ2∥Aix−v∥22+λ∥x∥1\frac{\rho}{2}\|A_ix-v\|_2^2+\lambda\|x\|_12ρ​∥Ai​x−v∥22​+λ∥x∥1​

over Rni\mathbb R^{n_i}Rni​ if and only if

∥AiTv∥∞=max⁡j∣(AiTv)j∣≤λ/ρ.\|A_i^Tv\|_\infty=\max_j\bigl|(A_i^Tv)_j\bigr|\le\lambda/\rho .∥AiT​v∥∞​=jmax​​(AiT​v)j​​≤λ/ρ.

The test lets a serial implementation skip the blocks whose features are not used.

Corrected statement. The book prints the test with the Euclidean norm, ∥AiT(Aixik+zˉk−Ax‾k−uk)∥2≤λ/ρ\|A_i^T(A_ix_i^k+\bar z^k-\overline{Ax}^k-u^k)\|_2\le\lambda/\rho∥AiT​(Ai​xik​+zˉk−Axk−uk)∥2​≤λ/ρ. That is the test for the group-lasso term λ∥xi∥2\lambda\|x_i\|_2λ∥xi​∥2​ of §8.3.2, and it is false for the ℓ1\ell_1ℓ1​ term: with Ai=I2A_i=I_2Ai​=I2​, ρ=λ=1\rho=\lambda=1ρ=λ=1 and v=(0.9,0.9)v=(0.9,0.9)v=(0.9,0.9) the minimizer is xi=0x_i=0xi​=0 although ∥v∥2≈1.27>1\|v\|_2\approx1.27>1∥v∥2​≈1.27>1. The correct test, stated here, uses ∥⋅∥∞\|\cdot\|_\infty∥⋅∥∞​, the dual norm of ∥⋅∥1\|\cdot\|_1∥⋅∥1​.

Formalization Note "xik+1=0x_i^{k+1}=0xik+1​=0" is read as "000 is a minimizer"; uniqueness of the minimizer is not claimed (the objective need not be strictly convex). λ\lambdaλ is lam in Lean.

Preamble
import Mathlib
import Definitions.Def_BoydADMM_ModelFit_Basic

open Matrix
Formal statement
namespace BoydADMM.ModelFit

/-- §8.3.1, p. 69, corrected: `0` minimizes `(ρ/2)‖Aᵢxᵢ − v‖₂² + λ‖xᵢ‖₁` iff
`‖Aᵢᵀv‖_∞ ≤ λ/ρ` (the book prints `‖·‖₂`, which is false for the ℓ1 block term). -/
theorem lasso_block_zero_test_corrected {m n : ℕ} (A : Matrix (Fin m) (Fin n) ℝ)
    (v : EuclideanSpace ℝ (Fin m)) (ρ lam : ℝ) (hρ : 0 < ρ) (hlam : 0 < lam) :
    IsMinOn (fun x : EuclideanSpace ℝ (Fin n) =>
        ρ / 2 * ‖Matrix.toEuclideanLin A x - v‖ ^ 2 + lam * l1norm x) Set.univ 0 ↔
      ∀ j, |Matrix.toEuclideanLin Aᵀ v j| ≤ lam / ρ := by sorry

end BoydADMM.ModelFit
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. 69, §8.3.1 (printed with ‖·‖₂; corrected to ‖·‖_∞)
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