Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 2 — for τ ∈ (0, 2/λmax(AᵀA)) the gradient step h = I − τg is nonexpansive, with equality iff h(X) − h(X′) = X − X′

Proved
MaGoldfarbFPC.Convergence.lemma2

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

nonexpansivep2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1proximal-gradient

Let A:Rm×n→Rp\mathcal A:\mathbb R^{m\times n}\to\mathbb R^pA:Rm×n→Rp be linear with matrix A∈Rp×mnA\in\mathbb R^{p\times mn}A∈Rp×mn, i.e. AX=A vec(X)\mathcal AX=A\,\mathrm{vec}(X)AX=Avec(X), let b∈Rpb\in\mathbb R^pb∈Rp, and let τ∈(0,2/λmax⁡(A⊤A))\tau\in(0,2/\lambda_{\max}(A^\top A))τ∈(0,2/λmax​(A⊤A)). Let g(X)=A∗(A(X)−b)g(X)=\mathcal A^*(\mathcal A(X)-b)g(X)=A∗(A(X)−b) and h(X)=X−τg(X)h(X)=X-\tau g(X)h(X)=X−τg(X). Then for all X,X′∈Rm×nX,X'\in\mathbb R^{m\times n}X,X′∈Rm×n

∥h(X)−h(X′)∥F≤∥X−X′∥F,\|h(X)-h(X')\|_F\le\|X-X'\|_F,∥h(X)−h(X′)∥F​≤∥X−X′∥F​,

and h(X)−h(X′)=X−X′h(X)-h(X')=X-X'h(X)−h(X′)=X−X′ if and only if ∥h(X)−h(X′)∥F=∥X−X′∥F\|h(X)-h(X')\|_F=\|X-X'\|_F∥h(X)−h(X′)∥F​=∥X−X′∥F​.

Together with Lemma 1 it makes the iteration map X↦Sτμ(h(X))X\mapsto S_{\tau\mu}(h(X))X↦Sτμ​(h(X)) non-expansive, with a usable description of when distances are preserved.

Formalization Note λmax⁡(A⊤A)=∥A∥22\lambda_{\max}(A^\top A)=\|\mathcal A\|_2^2λmax​(A⊤A)=∥A∥22​, and the range is the predicate StepRange τ Aop: 0<τ0<\tau0<τ and τ∥A∥22<2\tau\|\mathcal A\|_2^2<2τ∥A∥22​<2, which for A=0\mathcal A=0A=0 is (0,∞)(0,\infty)(0,∞) as on paper. "Let AX=A vec(X)\mathcal AX=A\,\mathrm{vec}(X)AX=Avec(X)" in the paper fixes notation for the matrix of A\mathcal AA; it is not a hypothesis. The statement holds for every bbb.

Preamble
import Mathlib
import Definitions.Def_MaGoldfarbFPC_Convergence_Basic
Formal statement
namespace MaGoldfarbFPC.Convergence

open CaiCandesShen.Convergence

/-- Lemma 2, p. 11: for `τ ∈ (0, 2/λ_max(AᵀA))` the operator `h(·) = I(·) − τg(·)` is
non-expansive in the Frobenius norm, and `h(X) − h(X′) = X − X′` if and only if
`‖h(X) − h(X′)‖_F = ‖X − X′‖_F`. -/
theorem lemma2 {m n p : ℕ} (Aop : Fin p → Mat m n) (b : Fin p → ℝ) (τ : ℝ)
    (hτ : StepRange τ Aop) (X X' : Mat m n) :
    frobNorm (h τ Aop b X - h τ Aop b X') ≤ frobNorm (X - X') ∧
      (h τ Aop b X - h τ Aop b X' = X - X' ↔
        frobNorm (h τ Aop b X - h τ Aop b X') = frobNorm (X - X')) := by sorry

end MaGoldfarbFPC.Convergence
Source
Ma, Goldfarb and Chen, Fixed point and Bregman iterative methods for matrix rank minimization, arXiv:0905.1643v2, p. 11, Lemma 2
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