Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proof of Theorem 5.1, p. 239 — the set N(x*, ε) = {‖x − x*‖ ≤ ε, ‖F(x)‖ ≤ ε/(2c + sp)} is invariant under one step

Proved
QiNonsmoothEq.Attraction.trapping_step

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

attractiondescent-methodnonsmooth-equationsp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1

Let F:Rn→RnF:\mathbb R^n\to\mathbb R^nF:Rn→Rn be locally Lipschitz and B-differentiable at a zero x∗x^*x∗, and let ccc satisfy ∥h∥≤c∥F′(x∗;h)∥\|h\|\le c\|F'(x^*;h)\|∥h∥≤c∥F′(x∗;h)∥ for all hhh (5.3). Let xk+1=xk+αkdkx^{k+1}=x^k+\alpha_kd^kxk+1=xk+αk​dk with αk≥0\alpha_k\ge 0αk​≥0 and

∥F(xk+1)∥≤∥F(xk)∥(5.1)\|F(x^{k+1})\|\le\|F(x^k)\| \tag{5.1}∥F(xk+1)∥≤∥F(xk)∥(5.1)

for all kkk, and let δ1,s,p>0\delta_1,s,p>0δ1​,s,p>0 be such that ∥xk−x∗∥≤δ1\|x^k-x^*\|\le\delta_1∥xk−x∗∥≤δ1​ implies αk≤s\alpha_k\le sαk​≤s and ∥dk∥≤p∥F(xk)∥\|d^k\|\le p\|F(x^k)\|∥dk∥≤p∥F(xk)∥ (5.2). Then there is δ∈(0,δ1)\delta\in(0,\delta_1)δ∈(0,δ1​) such that for every ϵ∈(0,δ)\epsilon\in(0,\delta)ϵ∈(0,δ) the set

N(x∗,ϵ)={x∈Rn: ∥x−x∗∥≤ϵ, ∥F(x)∥≤ϵ2c+sp}N(x^*,\epsilon)=\Big\{x\in\mathbb R^n:\ \|x-x^*\|\le\epsilon,\ \|F(x)\|\le\frac{\epsilon}{2c+sp}\Big\}N(x∗,ϵ)={x∈Rn: ∥x−x∗∥≤ϵ, ∥F(x)∥≤2c+spϵ​}

is invariant under one step of the method: for every kkk, xk∈N(x∗,ϵ)x^k\in N(x^*,\epsilon)xk∈N(x∗,ϵ) implies xk+1∈N(x∗,ϵ)x^{k+1}\in N(x^*,\epsilon)xk+1∈N(x∗,ϵ).

Once an iterate enters N(x∗,ϵ)N(x^*,\epsilon)N(x∗,ϵ), all later iterates stay there; since x∗x^*x∗ is a limiting point, every such set is entered, which gives convergence in Theorem 5.1.

Formalization Note The nonnegativity αk≥0\alpha_k\ge0αk​≥0 is an added hypothesis (see Theorem 5.1): the page's step ∥xk−x∗+αkdk∥≤∥xk−x∗∥+αk∥dk∥\|x^k-x^*+\alpha_kd^k\|\le\|x^k-x^*\|+\alpha_k\|d^k\|∥xk−x∗+αk​dk∥≤∥xk−x∗∥+αk​∥dk∥ uses it.

Preamble
import Mathlib
import Definitions.Def_NonsmoothNewton_Local_dirDeriv
import Definitions.Def_QiNonsmoothEq_Attraction_Setting
open Filter Topology NonsmoothNewton.Local
Formal statement
namespace QiNonsmoothEq.Attraction

/-- Qi 1993, proof of Theorem 5.1, p. 239: with `c` as in (5.3), there is `δ ∈ (0, δ₁)` such
that for every `ε ∈ (0, δ)` the set
`N(x*, ε) = {x : ‖x - x*‖ ≤ ε, ‖F(x)‖ ≤ ε / (2c + sp)}` is invariant under one step of the
descent direction method: `x^k ∈ N(x*, ε)` implies `x^{k+1} ∈ N(x*, ε)`. -/
theorem trapping_step {n : ℕ}
    (F : EuclideanSpace ℝ (Fin n) → EuclideanSpace ℝ (Fin n)) (hF : LocallyLipschitz F)
    (xstar : EuclideanSpace ℝ (Fin n)) (hBx : QiNonsmoothEq.Local.BDiffAt F xstar) (hzero : F xstar = 0)
    (c : ℝ) (hc : ∀ h : EuclideanSpace ℝ (Fin n), ‖h‖ ≤ c * ‖dirDeriv F xstar h‖)
    (x d : ℕ → EuclideanSpace ℝ (Fin n)) (α : ℕ → ℝ)
    (hstep : ∀ k, x (k + 1) = x k + α k • d k) (hα : ∀ k, 0 ≤ α k)
    (hdesc : ∀ k, ‖F (x (k + 1))‖ ≤ ‖F (x k)‖)
    (δ₁ s p : ℝ) (hδ₁ : 0 < δ₁) (hs : 0 < s) (hp : 0 < p)
    (hloc : ∀ k, ‖x k - xstar‖ ≤ δ₁ → α k ≤ s ∧ ‖d k‖ ≤ p * ‖F (x k)‖) :
    ∃ δ : ℝ, 0 < δ ∧ δ < δ₁ ∧ ∀ ε : ℝ, 0 < ε → ε < δ → ∀ k : ℕ,
      (‖x k - xstar‖ ≤ ε ∧ ‖F (x k)‖ ≤ ε / (2 * c + s * p)) →
      (‖x (k + 1) - xstar‖ ≤ ε ∧ ‖F (x (k + 1))‖ ≤ ε / (2 * c + s * p)) := by sorry

end QiNonsmoothEq.Attraction
Source
Qi, Convergence analysis of some algorithms for solving nonsmooth equations, Math. Oper. Res. 18 (1993), p. 239, proof of Theorem 5.1, definition of N(x*, ε) and the display after it
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