(7), p. 940 — Nahmias's identity y₊L = z + [max_l {v_l − d_[l,L)}]⁺
ProvedZipkinLostSales.Bounds.eq_7Let , let be a state, the current order, and the demands of the next periods. Write for the demand of periods through . Then the inventory on hand when the current order arrives, obtained by iterating the lost-sales dynamics, is
This closed form (due to Nahmias, 1979) expresses the only random quantity in the one-period cost through the state; in particular . It is the starting point for the bound on the last-period policy.
Formalization Note The identity is deterministic and is stated for all real , and demands, with no sign condition. is the model's yPlusL, defined by the on-hand recursion with , ; indices are , and the maximum is a Finset.sup' over the nonempty index set, nonempty because .
import Mathlib import Definitions.Def_ZipkinLostSales_Bounds_Model
namespace ZipkinLostSales.Bounds
/-- (7), p. 940 (Nahmias 1979): `y_{+L} = z + [max{v_l − d_{[l,L)} : l = 0, …, L − 1}]⁺`, where
`d_{[l,L)} = Σ_{l ≤ j < L} D j` is the demand of periods `t + l, …, t + L − 1`. A deterministic
identity, for every real `v`, `z` and `D`. -/
theorem eq_7 {L : ℕ} (hL : 0 < L) (v : Fin L → ℝ) (z : ℝ) (D : Fin L → ℝ) :
yPlusL v z D =
z + max ((Finset.univ : Finset (Fin L)).sup' ⟨⟨0, hL⟩, Finset.mem_univ _⟩
(fun l => v l - ∑ j ∈ Finset.Ici l, D j)) 0 := by sorry
end ZipkinLostSales.Bounds
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.