Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§8.2.1, p. 65 — the distributed-lasso xᵢ-update is ridge regression, xᵢ = (AᵢᵀAᵢ + ρI)⁻¹(Aᵢᵀbᵢ + ρ(z − uᵢ))

Proved
BoydADMM.ModelFit.lasso_x_update_ridge

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

admmlassop2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-bookp2o-v1ridge-regression

In distributed lasso by splitting across examples, processor iii holds a block Ai∈Rmi×nA_i\in\mathbb R^{m_i\times n}Ai​∈Rmi​×n of the feature matrix and the matching block bi∈Rmib_i\in\mathbb R^{m_i}bi​∈Rmi​ of the response, and its xix_ixi​-update minimizes a Tikhonov-regularized least-squares objective. Let ρ>0\rho>0ρ>0, z,ui∈Rnz,u_i\in\mathbb R^nz,ui​∈Rn (the current consensus variable zkz^kzk and scaled dual variable uiku_i^kuik​). Then a point x∈Rnx\in\mathbb R^nx∈Rn minimizes

12∥Aix−bi∥22+ρ2∥x−z+ui∥22\frac12\|A_ix-b_i\|_2^2+\frac{\rho}{2}\|x-z+u_i\|_2^221​∥Ai​x−bi​∥22​+2ρ​∥x−z+ui​∥22​

over Rn\mathbb R^nRn if and only if

x=(AiTAi+ρI)−1(AiTbi+ρ(z−ui)).x=(A_i^TA_i+\rho I)^{-1}\bigl(A_i^Tb_i+\rho(z-u_i)\bigr).x=(AiT​Ai​+ρI)−1(AiT​bi​+ρ(z−ui​)).

In particular this point is the unique minimizer.

This is the analytical solution the book gives for each local update of the distributed lasso, the step that runs in parallel over the data blocks.

Formalization Note The book writes "xik+1:=argmin⁡x_i^{k+1}:=\operatorname{argmin}xik+1​:=argmin … with analytical solution …"; the statement is the equivalence "minimizer   ⟺  \iff⟺ formula", which gives existence and uniqueness of the minimizer together. The iteration indices are dropped: z=zkz=z^kz=zk, ui=uiku_i=u_i^kui​=uik​.

Preamble
import Mathlib
import Definitions.Def_BoydADMM_ModelFit_Basic

open Matrix
Formal statement
namespace BoydADMM.ModelFit

/-- §8.2.1, p. 65: the distributed-lasso `xᵢ`-update is a ridge regression with the unique
solution `xᵢ = (AᵢᵀAᵢ + ρI)⁻¹(Aᵢᵀbᵢ + ρ(z − uᵢ))`. -/
theorem lasso_x_update_ridge {m n : ℕ} (A : Matrix (Fin m) (Fin n) ℝ)
    (b : EuclideanSpace ℝ (Fin m)) (z u : EuclideanSpace ℝ (Fin n)) (ρ : ℝ) (hρ : 0 < ρ)
    (x : EuclideanSpace ℝ (Fin n)) :
    IsMinOn (fun x' : EuclideanSpace ℝ (Fin n) =>
        1 / 2 * ‖Matrix.toEuclideanLin A x' - b‖ ^ 2 + ρ / 2 * ‖x' - z + u‖ ^ 2) Set.univ x ↔
      x = Matrix.toEuclideanLin ((Aᵀ * A + ρ • (1 : Matrix (Fin n) (Fin n) ℝ))⁻¹)
        (Matrix.toEuclideanLin Aᵀ b + ρ • (z - u)) := 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. 65, §8.2.1
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