Proposition 3.1(ii) — for convex , a WPV has a supporting hyperplane with nonnegative normal
ProvedNieYangPareto.proposition_3_1_iimoment-conemulti-objective-optimizationp2o-gran-per-chapterp2o-plan-paperp2o-v1pareto-optimalitypolynomial-optimization
Let and assume is convex. If is a weakly Pareto value, then there exists a supporting hyperplane for through whose normal is nonnegative: there exists such that
So on a convex epigraph set, weakly Pareto values are the values reached by nonnegative linear scalarizations.
Differences from the source:
- encoding: disjoint finite label sets ℰ, ℐ → two finite types E, I (
[Finite E] [Finite I]), each with its own constraint family; reason: separate types encode disjointness; the same instances - encoding: "a supporting hyperplane for 𝒰 through v whose normal is nonnegative, i.e., there exists 0 ≠ w ≥ 0 such that wᵀu ≥ wᵀv for all u ∈ 𝒰" → the "i.e." clause, verbatim; reason: the page states the two as equivalent, and v ∈ f(K) ⊆ 𝒰
Preamble
import Mathlib import Definitions.Def_NieYangPareto_MOP import Definitions.Def_NieYangPareto_ParetoNotions import Definitions.Def_NieYangPareto_MOP_U
Formal statement
namespace NieYangPareto
/-- **Proposition 3.1(ii).** Nie–Yang, arXiv:2108.04336v2, Proposition 3.1(ii), p.6: for `v ∈ f(K)`,
assume `𝒰` is convex. If `v` is a WPV, then there exists `0 ≠ w ≥ 0` such that `wᵀu ≥ wᵀv` for all
`u ∈ 𝒰`. -/
theorem proposition_3_1_ii {n m : ℕ} {E I : Type*} [Finite E] [Finite I]
(P : MOP n m E I)
(v : Fin m → ℝ) (hv : v ∈ P.obj '' P.K) (hU : Convex ℝ P.U)
(hWPV : P.IsWeaklyParetoValue v) :
∃ w : Fin m → ℝ, w ≠ 0 ∧ 0 ≤ w ∧ ∀ u ∈ P.U, w ⬝ᵥ v ≤ w ⬝ᵥ u := by sorry
end NieYangPareto
Source
Nie–Yang, The multi-objective polynomial optimization, arXiv:2108.04336v2, Proposition 3.1(ii), p.6
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.