Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proposition (1.4) — f°(x; ·) is the support function of ∂f(x)

Proved
ClarkeGradients.FlowInvariance.genDirDeriv_isGreatest

by mikedeng1 · Sep 27, 2026 · Mathlib 0df444a (Lean v4.33.1)

nonsmooth-analysisp2o-batch-p100bp2o-gran-per-chapterp2o-plan-paperp2o-v1support-function

Let f:Rn→Rf:\mathbb R^n\to\mathbb Rf:Rn→R be locally Lipschitz (Lipschitz on every bounded set), and let x,v∈Rnx,v\in\mathbb R^nx,v∈Rn. Then the generalized directional derivative (1.3) is the largest value of ζ⋅v\zeta\cdot vζ⋅v over the generalized gradient (1.1):

f∘(x;v)=max⁡{ζ⋅v: ζ∈∂f(x)}.f^\circ(x;v)=\max\{\zeta\cdot v:\ \zeta\in\partial f(x)\}.f∘(x;v)=max{ζ⋅v: ζ∈∂f(x)}.

That is, f∘(x;⋅)f^\circ(x;\cdot)f∘(x;⋅) is the support function of ∂f(x)\partial f(x)∂f(x), and the maximum is attained.

In this mission the proposition is applied to the distance function dFd_FdF​: it identifies dF∘(x;v)d_F^\circ(x;v)dF∘​(x;v) with a maximum over ∂dF(x)\partial d_F(x)∂dF​(x), which is how tangency of vvv is read off from dFd_FdF​.

Formalization Note "max" is encoded as IsGreatest of the image of ∂f(x)\partial f(x)∂f(x) under ζ↦⟨ζ,v⟩\zeta\mapsto\langle\zeta,v\rangleζ↦⟨ζ,v⟩, which asserts both that f∘(x;v)f^\circ(x;v)f∘(x;v) is an upper bound and that it is attained. The Lipschitz hypothesis is the §1 standing assumption and also guarantees that the real limsup in f∘f^\circf∘ is not a junk value. The same statement is a milestone of mission I.

Preamble
import Mathlib
import Definitions.Def_ClarkeGradients_Shared_LipschitzOnBounded
import Definitions.Def_ClarkeGradients_Shared_generalizedGradient
import Definitions.Def_ClarkeGradients_Shared_genDirDeriv
Formal statement
namespace ClarkeGradients.FlowInvariance

/-- Clarke (1975), Proposition (1.4): for locally Lipschitz `f`,
`f°(x; v) = max {ζ · v : ζ ∈ ∂f(x)}`, i.e. `f°(x; ·)` is the support function of `∂f(x)`. -/
theorem genDirDeriv_isGreatest {n : ℕ} (f : EuclideanSpace ℝ (Fin n) → ℝ)
    (hf : Shared.LipschitzOnBounded f) (x v : EuclideanSpace ℝ (Fin n)) :
    IsGreatest ((fun ζ => inner ℝ ζ v) '' Shared.generalizedGradient f x) (Shared.genDirDeriv f x v) := by sorry

end ClarkeGradients.FlowInvariance
Source
Clarke, Generalized gradients and applications, Trans. Amer. Math. Soc. 205 (1975), p. 248, Proposition (1.4)
Read-back

What the Lean code literally says, in plain math · claude-opus-5-5

The theorem takes:

  • n∈Nn \in \mathbb{N}n∈N;
  • a function f:Rn→Rf : \mathbb{R}^n \to \mathbb{R}f:Rn→R (Euclidean space) satisfying the imported hypothesis LipschitzOnBounded f, whose content is not part of this declaration's text;
  • arbitrary points x,v∈Rnx, v \in \mathbb{R}^nx,v∈Rn.

Let ∂f(x)\partial f(x)∂f(x) be the set produced by the imported operation generalizedGradient f x. Let f∘(x;v)f^\circ(x; v)f∘(x;v) be the real number produced by the imported operation genDirDeriv f x v. Neither definition is shown here. The theorem asserts that f∘(x;v)f^\circ(x;v)f∘(x;v) is the greatest element of the set {⟨ζ,v⟩:ζ∈∂f(x)}\{\langle \zeta, v\rangle : \zeta \in \partial f(x)\}{⟨ζ,v⟩:ζ∈∂f(x)} of inner products:

∃ ζ∗∈∂f(x): ⟨ζ∗,v⟩=f∘(x;v),and⟨ζ,v⟩≤f∘(x;v)  for all ζ∈∂f(x).\exists\, \zeta^\ast \in \partial f(x):\ \langle \zeta^\ast, v\rangle = f^\circ(x;v), \qquad\text{and}\qquad \langle \zeta, v\rangle \le f^\circ(x;v)\ \ \text{for all } \zeta \in \partial f(x).∃ζ∗∈∂f(x): ⟨ζ∗,v⟩=f∘(x;v),and⟨ζ,v⟩≤f∘(x;v)  for all ζ∈∂f(x).

In other words, the maximum is attained. In particular the theorem implies that ∂f(x)\partial f(x)∂f(x) is nonempty for every xxx.

Degenerate cases:

  • v=0v = 0v=0. The claim becomes f∘(x;0)=0f^\circ(x; 0) = 0f∘(x;0)=0, together with nonemptiness of ∂f(x)\partial f(x)∂f(x).
  • n=0n = 0n=0. v=0v = 0v=0 is forced, so the same conclusion applies.
Human review
  • Endorsed by Shuze Chen · Sep 28, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Sep 28, 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