Sect. 1, p. 4 (from [7]) and Sect. 5.1.1, p. 13 — a convex scenario program has at most d support constraints
ProvedWaitJudge.Convex.sstar_le_dimconvexityp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1scenario-optimizationsupport-constraints
Let and every be convex, let the tie-break functions be convex, and let Assumption 1 hold: every scenario program (9), for every sample size and every sample, has a unique tie-broken solution. Then for every and every sample the number of support constraints satisfies
The bound says that the number of support constraints never exceeds the number of optimization variables. It is what allows the probability of to be split into the events , , in (13).
Formalization Note The paper quotes this fact from Calafiore and Campi (2005). A support constraint is one whose removal changes the tie-broken solution. The statement is deterministic: no measure appears.
Preamble
import Mathlib import Definitions.Def_ScenarioApproach_Generalization_violation import Definitions.Def_WaitJudge_Convex_Setting open MeasureTheory ScenarioApproach.Generalization
Formal statement
namespace WaitJudge.Convex
theorem sstar_le_dim {d p : ℕ} {Δ : Type*} (c : E d) (X : Set (E d)) (Xδ : Δ → Set (E d))
(tb : Fin p → E d → ℝ) (hX : Convex ℝ X) (hXδ : ∀ δ, Convex ℝ (Xδ δ))
(htb : ∀ j, ConvexOn ℝ Set.univ (tb j)) (hA1 : Assumption1 c X Xδ tb) :
∀ (m : ℕ) (ω : Fin m → Δ), sstar c X Xδ tb ω ≤ d := by sorry
end WaitJudge.Convex
Source
Campi & Garatti, Wait-and-judge scenario optimization, Math. Program. (2018), doi:10.1007/s10107-016-1056-9 (accepted manuscript), PDF p. 4 (paragraph after Definition 2, citing [7]) and PDF p. 13 (before (13))
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.