Eq. (3.1), p. 23 — z_Stoch(b, d) ≤ E_µ[min(d₁(ω), …, dₙ(ω))] in the instance of Theorem 3.1
ProvedRobustPower.CostGap.eq_3_1_stoch_le_expected_minConsider the instance of Theorem 3.1 (, , , continuous second-stage variables, , ), with a probability measure on the scenario set and a measurable cost map whose range is exactly . Then the optimal value of the stochastic problem satisfies
The paper obtains this bound from a policy that, in each scenario, buys one unit of a cheapest coordinate. No distributional assumption on is used beyond measurability and boundedness; combined with Eq. (3.2) it gives .
Formalization Note is the extended-real infimum of (1.4) over integrable policies. The paper's policy puts a 1 on every minimizing coordinate, which at ties costs a multiple of the minimum; the statement here is the resulting bound on , not a claim about that particular policy. The minimum is written as an infimum over the finite nonempty index set .
import Mathlib import Definitions.Def_RobustPower_CostGap_Problems open MeasureTheory Matrix
namespace RobustPower.CostGap
/-- Eq. (3.1), p. 23: in the instance of Theorem 3.1, the policy that buys one unit of a cheapest
coordinate shows `z_Stoch(b, d) ≤ E_μ[min(d₁(ω), …, dₙ(ω))]`. -/
theorem eq_3_1_stoch_le_expected_min {n : ℕ} (hn : 0 < n)
{Ω : Type*} [MeasurableSpace Ω] (μ : Measure Ω) [IsProbabilityMeasure μ]
(d : Ω → Fin n → ℝ) (hd : Measurable d) (hrange : Set.range d = Set.Icc 0 1) :
zStochBD μ (0 : Matrix (Fin 1) (Fin 0) ℝ) (fun _ _ => 1 : Matrix (Fin 1) (Fin n) ℝ)
(fun _ _ => 1 : Ω → Fin 1 → ℝ) (0 : Fin 0 → ℝ) d ∅ ∅ ≤
((∫ ω, (⨅ j, d ω j) ∂μ : ℝ) : EReal) := by sorry
end RobustPower.CostGap
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.