Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 9, p. 20 — f, h semiconvex and 0 ∈ M(x̄): no feasible point, or x̄ optimal, or no strictly feasible point

Proved
MifflinSemismooth.Optimality.theorem9

by mikedeng1 · Oct 8, 2026 · Mathlib 0df444a (Lean v4.33.1)

nonsmooth-optimizationp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1semiconvexstationaritysufficient-optimality

Consider the problem of minimizing f(x)f(x)f(x) subject to h(x)≤0h(x)\le 0h(x)≤0, with f,h:Rn→Rf,h:\mathbb R^n\to\mathbb Rf,h:Rn→R, and the map MMM of §5 (M(x)=∂f(x)M(x)=\partial f(x)M(x)=∂f(x) if h(x)<0h(x)<0h(x)<0, conv⁡{∂f(x)∪∂h(x)}\operatorname{conv}\{\partial f(x)\cup\partial h(x)\}conv{∂f(x)∪∂h(x)} if h(x)=0h(x)=0h(x)=0, ∂h(x)\partial h(x)∂h(x) if h(x)>0h(x)>0h(x)>0). Suppose fff and hhh are semiconvex on Rn\mathbb R^nRn and xˉ∈Rn\bar x\in\mathbb R^nxˉ∈Rn satisfies 0∈M(xˉ)0\in M(\bar x)0∈M(xˉ).

  1. If h(xˉ)>0h(\bar x)>0h(xˉ)>0, then h(x)≥h(xˉ)>0h(x)\ge h(\bar x)>0h(x)≥h(xˉ)>0 for all x∈Rnx\in\mathbb R^nx∈Rn; that is, the problem has no feasible points.
  2. If h(xˉ)≤0h(\bar x)\le 0h(xˉ)≤0, then at least one of the following holds:
    • (i) xˉ\bar xxˉ is optimal;
    • (ii) h(x)≥0h(x)\ge 0h(x)≥0 for all x∈Rnx\in\mathbb R^nx∈Rn; that is, the problem has no strictly feasible points.

In formulas:

(h(xˉ)>0⇒∀x, h(x)≥h(xˉ)>0) ∧ (h(xˉ)≤0⇒xˉ optimal ∨ ∀x, h(x)≥0).\bigl(h(\bar x)>0\Rightarrow \forall x,\ h(x)\ge h(\bar x)>0\bigr)\ \wedge\ \bigl(h(\bar x)\le 0\Rightarrow \bar x\ \text{optimal}\ \vee\ \forall x,\ h(x)\ge 0\bigr).(h(xˉ)>0⇒∀x, h(x)≥h(xˉ)>0) ∧ (h(xˉ)≤0⇒xˉ optimal ∨ ∀x, h(x)≥0).

Together with Theorem 7 (optimality implies stationarity) this says that for semiconvex problem functions with a strictly feasible point (a Slater-type condition), stationarity is equivalent to optimality. It is the nonsmooth analogue of the classical sufficiency of the Karush–Kuhn–Tucker conditions for pseudoconvex objective and quasiconvex constraints.

Formalization Note "Semiconvex on Rn\mathbb R^nRn" is semiconvexity at each point with respect to X=RnX=\mathbb R^nX=Rn. The hypothesis is 0∈M(xˉ)0\in M(\bar x)0∈M(xˉ) alone, without feasibility, since case 1 concerns infeasible xˉ\bar xxˉ. The theorem is stated for an arbitrary constraint function hhh; the paper's h=max⁡ihih=\max_i h_ih=maxi​hi​ is a special case, and Theorem 6 transfers semiconvexity from the hih_ihi​ to hhh.

Preamble
import Mathlib
import Definitions.Def_MifflinSemismooth_Optimality_Setting
Formal statement
namespace MifflinSemismooth.Optimality

/-- Mifflin (1976), §5, Theorem 9, p. 20: if `f` and `h` are semiconvex on `ℝⁿ` and `0 ∈ M(x̄)`,
then (a) `h(x̄) > 0` implies `h(x) ≥ h(x̄) > 0` for all `x`; (b) `h(x̄) ≤ 0` implies that `x̄` is
optimal or `h(x) ≥ 0` for all `x`. -/
theorem theorem9 {n : ℕ} (f h : EuclideanSpace ℝ (Fin n) → ℝ) (hf : MifflinSemismooth.Extremal.SemiconvexOn Set.univ f)
    (hh : MifflinSemismooth.Extremal.SemiconvexOn Set.univ h) (xbar : EuclideanSpace ℝ (Fin n))
    (h0 : (0 : EuclideanSpace ℝ (Fin n)) ∈ Mmap f h xbar) :
    (0 < h xbar → ∀ x, h xbar ≤ h x ∧ 0 < h x) ∧
      (h xbar ≤ 0 → IsOptimal f h xbar ∨ ∀ x, 0 ≤ h x) := by sorry

end MifflinSemismooth.Optimality
Source
Mifflin, Semismooth and semiconvex functions in constrained optimization, IIASA Research Report RR-76-21 (December 1976), p. 20, Theorem 9
Human review
  • Endorsed by Shuze Chen · Oct 8, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 8, 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