Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Eq. (3.4), p. 267, in an arbitrary norm (p. 272) — 0 ≤ f(x) − f(y) − ∇f(y)⊤(x − y) ≤ (β/2)‖x − y‖²

Proved
ConvexOptAlg.FrankWolfe.eq_3_4

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

convex-optimizationp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-bookp2o-v1smoothness

Let EEE be a finite-dimensional real normed space, X⊆E\mathcal X\subseteq EX⊆E convex, and f:E→Rf:E\to\mathbb Rf:E→R differentiable, convex on X\mathcal XX, and β\betaβ-smooth with respect to ∥⋅∥\|\cdot\|∥⋅∥ on X\mathcal XX, i.e. ∥∇f(x)−∇f(y)∥∗≤β∥x−y∥\|\nabla f(x)-\nabla f(y)\|_*\le\beta\|x-y\|∥∇f(x)−∇f(y)∥∗​≤β∥x−y∥ for x,y∈Xx,y\in\mathcal Xx,y∈X, where ∥⋅∥∗\|\cdot\|_*∥⋅∥∗​ is the dual norm. Then for all x,y∈Xx,y\in\mathcal Xx,y∈X,

0≤f(x)−f(y)−∇f(y)⊤(x−y)≤β2∥x−y∥2.0\le f(x)-f(y)-\nabla f(y)^\top(x-y)\le\frac{\beta}{2}\|x-y\|^2 .0≤f(x)−f(y)−∇f(y)⊤(x−y)≤2β​∥x−y∥2.

This is inequality (3.4) of Section 3.2, which the book proves for the Euclidean norm and then uses in the proof of Theorem 3.8 with the remark that it holds for smoothness in an arbitrary norm. Its upper half is the first step of the one-step analysis of conditional gradient descent.

Formalization Note ∇f(y)⊤v\nabla f(y)^\top v∇f(y)⊤v is f' y v for a derivative map f' with HasFDerivAt f (f' y) y, and ∥⋅∥∗\|\cdot\|_*∥⋅∥∗​ is the operator norm. Convexity and smoothness are assumed on the convex set X\mathcal XX and the inequality is stated for x,y∈Xx,y\in\mathcal Xx,y∈X; the book states (3.4) for all of Rn\mathbb R^nRn, which is the case X=E\mathcal X=EX=E.

Preamble
import Mathlib
import Definitions.Def_ConvexOptAlg_FrankWolfe_Defs
Formal statement
namespace ConvexOptAlg.FrankWolfe

/-- Bubeck, arXiv:1405.4980v2, Eq. (3.4), p. 267, in an arbitrary norm as used in the proof of
Theorem 3.8, p. 272 ("it can easily be seen that (3.4) holds true for smoothness in an arbitrary
norm"): if `f` is convex and β-smooth w.r.t. `‖·‖` on the convex set `X`, then for `x, y ∈ X`,
`0 ≤ f(x) − f(y) − ∇f(y)⊤(x − y) ≤ (β/2)‖x − y‖²`. The pairing `∇f(y)⊤v` is `f' y v` and the
dual norm is the operator norm. -/
theorem eq_3_4 {E : Type*} [NormedAddCommGroup E] [NormedSpace ℝ E] [FiniteDimensional ℝ E]
    (X : Set E) (hXconv : Convex ℝ X)
    (f : E → ℝ) (f' : E → E →L[ℝ] ℝ) (β : ℝ)
    (hf : ConvexOn ℝ X f) (hsmooth : IsBetaSmoothNormOn X f f' β)
    (x y : E) (hx : x ∈ X) (hy : y ∈ X) :
    0 ≤ f x - f y - f' y (x - y) ∧ f x - f y - f' y (x - y) ≤ β / 2 * ‖x - y‖ ^ 2 := by sorry

end ConvexOptAlg.FrankWolfe
Source
Bubeck, arXiv:1405.4980v2, Eq. (3.4), p. 267; proof of Theorem 3.8, p. 272 ("it can easily be seen that (3.4) holds true for smoothness in an arbitrary norm")
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