Proposition 3.1, p. 12 — Π_{1,α}(q) = m₁ − q(m₁^α/m_α)^{1/(α−1)} for 0 ≤ q ≤ q₀, attained by the two-point law (3.6)
ProvedHeavyTailNV.ZeroOrder.proposition_3_1distributionally-robust-optimizationmoment-problemnewsvendorp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Let and , and let be the set of laws on with mean and -th moment . Write and let be the two-point law
Then , and for every with ,
For small order quantities the worst-case expected shortage is linear in , and the same two-point law is worst-case for every such . This closed form drives the zero-order threshold of Proposition 3.2.
Formalization Note "The worst-case demand distribution is given as (3.6)" is stated as two facts: the law lies in , and its expected shortage equals the supremum for every .
Preamble
import Mathlib import Definitions.Def_HeavyTailNV_ZeroOrder_Model
Formal statement
namespace HeavyTailNV.ZeroOrder
open MeasureTheory
/-- Proposition 3.1 (arXiv:1806.05379v2, p. 12): for `0 ≤ q ≤ q₀` the worst-case expected
shortage over `F_{1,α}` is `m₁ - q (m₁^α/m_α)^{1/(α-1)}` (3.5), and the two-point law (3.6) lies in
`F_{1,α}` and attains it. -/
theorem proposition_3_1 (α m1 ma : ℝ) (hα : 1 < α) (hm1 : 0 < m1) (hma : m1 ^ α < ma) :
twoPointLaw m1 ma α ∈ HeavyTailNV.Tail.ambiguitySet m1 ma α ∧
∀ q : ℝ, 0 ≤ q → q ≤ q0 m1 ma α →
HeavyTailNV.Tail.worstCase m1 ma α q = m1 - q * (m1 ^ α / ma) ^ (1 / (α - 1)) ∧
∫ w, max (w - q) 0 ∂(twoPointLaw m1 ma α) = HeavyTailNV.Tail.worstCase m1 ma α q := by sorry
end HeavyTailNV.ZeroOrder
Source
Das, Dhara and Natarajan, On the heavy-tail behavior of the distributionally robust newsvendor, arXiv:1806.05379v2, p. 12, Proposition 3.1, (3.5)–(3.6)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.