Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 4.2, p. 9 — step (4.ii) of the adaptive projected reflected gradient method is well-defined

Proved
ProjReflGrad.Adaptive.lemma_4_2

by mikedeng1 · Oct 9, 2026 · Mathlib 0df444a (Lean v4.33.1)

adaptive-step-sizep2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1variational-inequality

Let F:H→HF : H \to HF:H→H be LLL-Lipschitz with L>0L > 0L>0, let α>0\alpha > 0α>0, and let the previous data of Algorithm 4.2 be points xn,xn−1,yn−1x_n, x_{n-1}, y_{n-1}xn​,xn−1​,yn−1​, a weight τn−1≥0\tau_{n-1} \ge 0τn−1​≥0 and a step λn−1\lambda_{n-1}λn−1​ with 0<λn−1≤λˉ0 < \lambda_{n-1} \le \bar\lambda0<λn−1​≤λˉ. For τ∈(0,1]\tau \in (0,1]τ∈(0,1] write y′(τ)=xn+τ(xn−xn−1)y'(\tau) = x_n + \tau(x_n - x_{n-1})y′(τ)=xn​+τ(xn​−xn−1​) and let λ(⋅,⋅)\lambda(\cdot,\cdot)λ(⋅,⋅) be the step rule (4.1). Then there is τn′∈(0,1]\tau'_n \in (0, 1]τn′​∈(0,1] with

λ(y′(τn′),τn′)≥τn′λn−1(4.3)\lambda(y'(\tau'_n), \tau'_n) \ge \tau'_n\lambda_{n-1} \qquad (4.3)λ(y′(τn′​),τn′​)≥τn′​λn−1​(4.3)

and there is λn′∈[τn′λn−1,λ(y′(τn′),τn′)]\lambda'_n \in [\tau'_n\lambda_{n-1}, \lambda(y'(\tau'_n), \tau'_n)]λn′​∈[τn′​λn−1​,λ(y′(τn′​),τn′​)] with

∥λn′F(y′(τn′))−τn′λn−1F(yn−1)∥≤α∥y′(τn′)−yn−1∥.(4.4)\|\lambda'_n F(y'(\tau'_n)) - \tau'_n\lambda_{n-1}F(y_{n-1})\| \le \alpha\|y'(\tau'_n) - y_{n-1}\|. \qquad (4.4)∥λn′​F(y′(τn′​))−τn′​λn−1​F(yn−1​)∥≤α∥y′(τn′​)−yn−1​∥.(4.4)

This is Lemma 4.2: the choices required in step (4.ii) of Algorithm 4.2 are always possible. Along every run the hypotheses 0<λn−1≤λˉ0 < \lambda_{n-1} \le \bar\lambda0<λn−1​≤λˉ and τn−1∈(0,1]\tau_{n-1} \in (0,1]τn−1​∈(0,1] hold, so the lemma applies at every step.

Formalization Note The statement does not assume λˉ≥α/L\bar\lambda \ge \alpha/Lλˉ≥α/L or λn−1≥α/L\lambda_{n-1} \ge \alpha/Lλn−1​≥α/L. The paper's proof takes τn′=α/(λn−1L)\tau'_n = \alpha/(\lambda_{n-1}L)τn′​=α/(λn−1​L), which lies in (0,1](0,1](0,1] only when λn−1≥α/L\lambda_{n-1} \ge \alpha/Lλn−1​≥α/L; the statement as printed holds without that bound. It also does not assume the branch condition λ^n<λn−1\hat\lambda_n < \lambda_{n-1}λ^n​<λn−1​, which the conclusion does not need.

Preamble
import Mathlib
import Definitions.Def_ProjReflGrad_Adaptive_Setting
Formal statement
namespace ProjReflGrad.Adaptive

/-- Lemma 4.2 (Malitsky 2015, p. 9): step (4.ii) of Algorithm 4.2 is well-defined. For an
`L`-Lipschitz `F`, `α > 0` and previous data `0 < λ_{n-1} ≤ λ̄`, `τ_{n-1} ≥ 0`, some
`τ'_n ∈ (0, 1]` satisfies (4.3) for `y'_n = x_n + τ'_n (x_n - x_{n-1})`, and then some
`λ'_n ∈ [τ'_n λ_{n-1}, λ(y'_n, τ'_n)]` satisfies (4.4). -/
theorem lemma_4_2 {H : Type*} [NormedAddCommGroup H] [InnerProductSpace ℝ H] (F : H → H) (L : ℝ) (hL : 0 < L)
    (hC3 : ProjReflGrad.Weak.IsLipschitzMap F L) (α lamBar : ℝ) (hα : 0 < α) (xn xp yp : H) (taup lamp : ℝ)
    (htaup : 0 ≤ taup) (hlamp : 0 < lamp) (hlampBar : lamp ≤ lamBar) :
    ∃ τ ∈ Set.Ioc (0 : ℝ) 1,
      τ * lamp ≤ stepCap F α lamBar yp taup lamp (xn + τ • (xn - xp)) τ ∧
      ∃ l ∈ Set.Icc (τ * lamp) (stepCap F α lamBar yp taup lamp (xn + τ • (xn - xp)) τ),
        ‖l • F (xn + τ • (xn - xp)) - (τ * lamp) • F yp‖
          ≤ α * ‖(xn + τ • (xn - xp)) - yp‖ := by sorry

end ProjReflGrad.Adaptive
Source
Malitsky, Projected Reflected Gradient Methods for Monotone Variational Inequalities, arXiv:1502.04968v1, p. 9, Lemma 4.2
Human review
  • Endorsed by Shuze Chen · Oct 9, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 9, 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