Lemma 6, p. 940 — the last-period policy z̄_T(v) is in Z(s̄)
ProvedZipkinLostSales.Bounds.lemma_6Consider the lost-sales model of §4 under its standing assumptions (i.i.d. nonnegative demand with finite mean, , , with ). In the last period the function to be minimized is the one-period cost . For every state , the smallest minimizer of over exists and satisfies the bounds of :
This is the base case of the bounds: the myopic order is limited by the fractiles (6) of the lead-time demand.
Formalization Note is periods of continuation, where (with ). "The policy " is read as the least minimizer, and its existence is part of the statement. The statement is about §4's recursion; its identification with §2's optimal policy ("some algebra", p. 940) is not formalized. The finite mean and the sign conventions on are added readings of §2.
import Mathlib import Definitions.Def_ZipkinLostSales_Bounds_Model
namespace ZipkinLostSales.Bounds
/-- Lemma 6, p. 940: the policy `z̄_T(v)` is in `Z(s̄)`. With one period to go (`k = 0`,
`t = T`), for every `v ∈ ZipkinLostSales.LNatural.V` the smallest minimizer of `g_T(v, ·) = q(v, ·)` over `z ≥ 0` exists and
satisfies the bounds of `Z(s̄)` at `v`. -/
theorem lemma_6 {L : ℕ} (M : Data) (hM : Assumptions L M) (v : Fin L → ℝ) (hv : v ∈ ZipkinLostSales.LNatural.V L) :
∃ z : ℝ, IsOptOrder M 0 v z ∧ InZ L M v z := by sorry
end ZipkinLostSales.Bounds
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.