Theorem 3 — pseudoconvexity through the Clarke subdifferential
OpenTiltedCvx.Supplement.theorem_3convex-analysisgeneralized-convexitynonsmooth-analysisoptimizationp2o-gran-per-chapterp2o-plan-bookp2o-strengthenedp2o-v1
Let be proper, lower semicontinuous and radially continuous. The following are equivalent:
- for all and with , ;
- is –tilted convex;
- is pseudoconvex (Definition 10);
- the operator is –tilted monotone;
- the operator is pseudomonotone.
With for and for , the endpoint of the tilted scale is pseudoconvexity, whereas is quasiconvexity (Theorem 2).
Formalization Note Pseudoconvexity is the general Definition 10 with the Clarke subdifferential, so items 1 and 3 coincide by definition, as in the paper's proof.
Differences from the source:
- generalization: functions on ℝ^d → functions on a finite-dimensional real inner product space E; reason: every such E is isometric to ℝ^d and ℝ^d = EuclideanSpace ℝ (Fin d) is an instance, so the statement specializes to the paper's
- encoding: ℝ̄ = ℝ ∪ {±∞}, dom(f) = {f finite} → EReal and the published TiltedCvx.effDom; reason: Lean convention; same instances
- encoding: pseudoconvexity of a proper lsc radially continuous f (Definition 10) → IsPseudoconvex f, whose predicate omits the standing hypotheses; the theorem carries them; reason: the theorem assumes exactly those hypotheses, so the predicate is the paper's on every f it is applied to
Preamble
import Mathlib import Definitions.Def_TiltedCvx_effDom import Definitions.Def_TiltedCvx_IsProper import Definitions.Def_ClarkeStrat_Proj_Setting import Definitions.Def_TiltedCvx_Supplement_RadiallyContinuous import Definitions.Def_TiltedCvx_Supplement_IsPseudoconvex import Definitions.Def_TiltedCvx_Supplement_IsPseudomonotone import Definitions.Def_TiltedCvx_Supplement_TiltedConvexZeroPlus import Definitions.Def_TiltedCvx_Supplement_TiltedMonotoneZeroPlus open scoped InnerProductSpace open ClarkeStrat.Proj
Formal statement
namespace TiltedCvx.Supplement
/-- **Theorem 3 (pseudoconvexity through the Clarke subdifferential).**
Pedramfar–Khanna–Quinn–Aggarwal, arXiv:2611.03472, Theorem 3, p.17. Let `f : ℝ^d → ℝ̄` be a
proper, lsc, and radially continuous function. Then the following are equivalent:
(i) for all `x, y ∈ dom(f)` and `v ∈ ∂̄f(x)` such that `⟨y − x, v⟩ ≥ 0`, `f(y) ≥ f(x)`;
(ii) `f` is 0⁺–tilted convex;
(iii) `f` is pseudoconvex;
(iv) the operator `∂̄f` is 0⁺–tilted monotone;
(v) the operator `∂̄f` is pseudomonotone.
**Formalization Note** Pseudoconvexity is the general Definition 10 (with the Clarke `∂̄f`), so
(i) and (iii) coincide by definition, as the paper's proof notes. -/
theorem theorem_3 {E : Type*} [NormedAddCommGroup E] [InnerProductSpace ℝ E]
[FiniteDimensional ℝ E] (f : E → EReal) (hf : TiltedCvx.IsProper f)
(hlsc : LowerSemicontinuous f) (hrad : RadiallyContinuous f) :
List.TFAE
[∀ x ∈ TiltedCvx.effDom f, ∀ y ∈ TiltedCvx.effDom f, ∀ v ∈ ClarkeSubdiff f x,
0 ≤ ⟪y - x, v⟫_ℝ → f x ≤ f y,
TiltedConvexZeroPlus f,
IsPseudoconvex f,
TiltedMonotoneZeroPlus (ClarkeSubdiff f),
IsPseudomonotone (ClarkeSubdiff f)] := by sorry
end TiltedCvx.Supplement
Source
Pedramfar–Khanna–Quinn–Aggarwal, Tilted Convexity and an Exact Offline–Online Separation, arXiv:2611.03472, Theorem 3, p.17