Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proposition 5.1(a), p. 173 — ψ_e is 1-strongly convex on int Δ w.r.t. ‖·‖₁: Σ(x_j − y_j) ln(x_j/y_j) ≥ ‖x − y‖₁²

Proved
BeckTeboulleMD.EMDA.proposition_5_1_a

by mikedeng1 · Oct 5, 2026 · Mathlib 0df444a (Lean v4.33.1)

entropyp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1pinsker-inequalitystrong-convexityunit-simplex

Let ψe(x)=∑j=1nxjln⁡xj\psi_e(x) = \sum_{j=1}^n x_j \ln x_jψe​(x)=∑j=1n​xj​lnxj​ be the entropy and int⁡Δ={x∈Rn:xj>0, ∑jxj=1}\operatorname{int}\Delta = \{x \in \mathbb R^n : x_j > 0,\ \sum_j x_j = 1\}intΔ={x∈Rn:xj​>0, ∑j​xj​=1}. Then for all x,y∈int⁡Δx, y \in \operatorname{int}\Deltax,y∈intΔ:

  1. ⟨∇ψe(x)−∇ψe(y), x−y⟩=∑j=1n(xj−yj)ln⁡xjyj\langle \nabla\psi_e(x) - \nabla\psi_e(y),\ x - y\rangle = \sum_{j=1}^n (x_j - y_j)\ln\dfrac{x_j}{y_j}⟨∇ψe​(x)−∇ψe​(y), x−y⟩=∑j=1n​(xj​−yj​)lnyj​xj​​;
∑j=1n(xj−yj)ln⁡xjyj≥∥x−y∥12;\sum_{j=1}^n (x_j - y_j)\ln\frac{x_j}{y_j} \ge \|x - y\|_1^2;j=1∑n​(xj​−yj​)lnyj​xj​​≥∥x−y∥12​;
  1. ψe\psi_eψe​ is 111-strongly convex over int⁡Δ\operatorname{int}\DeltaintΔ with respect to ∥⋅∥1\|\cdot\|_1∥⋅∥1​: for a,b≥0a, b \ge 0a,b≥0 with a+b=1a + b = 1a+b=1,
ψe(ax+by)≤aψe(x)+bψe(y)−12 ab ∥x−y∥12.\psi_e(a x + b y) \le a\psi_e(x) + b\psi_e(y) - \tfrac12\, a b\, \|x - y\|_1^2.ψe​(ax+by)≤aψe​(x)+bψe​(y)−21​ab∥x−y∥12​.

The strong convexity constant σ=1\sigma = 1σ=1 with respect to the ℓ1\ell_1ℓ1​ norm is what makes the entropic mirror descent bound depend on the dimension only through ln⁡n\ln nlnn.

Formalization Note ∥x−y∥1\|x - y\|_1∥x−y∥1​ is written ∑j∣xj−yj∣\sum_j |x_j - y_j|∑j​∣xj​−yj​∣, and the strong convexity is written out explicitly instead of through a normed space carrying the ℓ1\ell_1ℓ1​ norm; it is Mathlib's notion of strong convexity with parameter 111 for that norm. ∇ψe(x)\nabla\psi_e(x)∇ψe​(x) is the Fréchet derivative of ψe\psi_eψe​ on Rn\mathbb R^nRn (which does not depend on the norm). "int⁡Δ\operatorname{int}\DeltaintΔ" is the relative interior of Δ\DeltaΔ, since Δ\DeltaΔ has empty interior in Rn\mathbb R^nRn.

Preamble
import Mathlib
import Definitions.Def_BeckTeboulleMD_EMDA_Setting
Formal statement
namespace BeckTeboulleMD.EMDA

/-- Proposition 5.1(a), p. 173: on the relative interior of the unit simplex,
`⟨∇ψ_e(x) − ∇ψ_e(y), x − y⟩ = ∑_j (x_j − y_j) ln(x_j / y_j) ≥ ‖x − y‖₁²`,
and `ψ_e` is `1`-strongly convex there with respect to `‖·‖₁`. -/
theorem proposition_5_1_a {n : ℕ} :
    (∀ x ∈ relIntSimplex n, ∀ y ∈ relIntSimplex n,
      (fderiv ℝ (entropy (n := n)) x - fderiv ℝ (entropy (n := n)) y) (x - y)
        = ∑ j, (x j - y j) * Real.log (x j / y j)) ∧
    (∀ x ∈ relIntSimplex n, ∀ y ∈ relIntSimplex n,
      (∑ j, |x j - y j|) ^ 2 ≤ ∑ j, (x j - y j) * Real.log (x j / y j)) ∧
    (∀ x ∈ relIntSimplex n, ∀ y ∈ relIntSimplex n, ∀ a b : ℝ, 0 ≤ a → 0 ≤ b → a + b = 1 →
      entropy (a • x + b • y)
        ≤ a * entropy x + b * entropy y - a * b * (1 / 2 * (∑ j, |x j - y j|) ^ 2)) := by sorry

end BeckTeboulleMD.EMDA
Source
Beck & Teboulle, Mirror descent and nonlinear projected subgradient methods for convex optimization, Oper. Res. Lett. 31 (2003), p. 173, Proposition 5.1(a)
Human review
  • Endorsed by Shuze Chen · Oct 5, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 5, 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