§2.4 — for 0 ≤ ŷ ≤ y, Q(y) − Q(ŷ) ≤ h·(y − ŷ)
ProvedAIMInventory.Known.cost_increment_leinventorynewsvendorp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1
Let demand have law with finite mean, let , and let . For all levels ,
In the non-perishable case the AIM algorithm implements ; this inequality turns the excess cost into times the excess inventory , which Theorem 6 bounds by a queue.
Preamble
import Mathlib import Definitions.Def_AIMInventory_Known_AIM
Formal statement
namespace AIMInventory.Known
/-- §2.4, p. 10: raising the order-up-to level from `ŷ` to `y ≥ ŷ ≥ 0` raises the expected
one-period cost by at most `h·(y − ŷ)`. -/
theorem cost_increment_le (P : CappedBaseStock.DemandLaw) (h b : ℝ) (hh : 0 < h) (hb : 0 < b)
(yhat y : ℝ) (hyhat : 0 ≤ yhat) (hle : yhat ≤ y) :
Q P h b y - Q P h b yhat ≤ h * (y - yhat) := by sorry
end AIMInventory.Known
Source
Huh & Rusmevichientong, A Non-Parametric Asymptotic Analysis of Inventory Planning with Censored Demand, Math. Oper. Res. (2009), author's manuscript, p. 10, §2.4 (display after 'Since y_t = max{ŷ_t, x_t}')
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.