Theorem 4, p. 939 — for all t, the lost-sales cost functions f̄_t(v) and ḡ_t(v, ζ) are L♮-convex
ProvedZipkinLostSales.LNatural.theorem_4Theorem 4 (Karlin and Scarf 1958 and Morton 1969). Consider the standard single-item lost-sales inventory system in discrete time: lead time ; independent nonnegative demands with a common law of finite mean; unit procurement cost , holding cost , lost-sales penalty ; discount factor . In the transformed state (partial sums of the pipeline, ) and transformed order , let
with and .
Then for all , the functions and are L♮-convex: is submodular in on , and is submodular in for , .
L♮-convexity is the structural property from which the paper derives that the optimal order is monotone in the transformed state with limited sensitivity (Corollary 5), and from which convexity of follows.
Formalization Note. Data are stationary, so depends on only through the number of periods to go ; the statement is for every , pairing fbar … k with gbar … k (whose continuation is fbar … k), which covers every of every horizon . Vectors are indexed with . The paper's "min" over is an infimum; under the hypotheses all costs are nonnegative, so it is not Lean's junk value. The cost hypotheses read the paper's "unit cost" and "discount rate"; nonnegative independent demand is the paper's, with one law because data are stationary; finite mean demand is added so that is finite. Expectations are Bochner integrals.
import Mathlib import Definitions.Def_ZipkinLostSales_LNatural_Model open MeasureTheory
namespace ZipkinLostSales.LNatural
/-- Theorem 4 (Karlin and Scarf 1958 and Morton 1969; Zipkin 2008, p. 939): for all `t`, the
functions `f̄_t(v)` and `ḡ_t(v, ζ)` are L♮-convex. With `k` the number of periods to go,
`fbar … k` is L♮-convex on `V` and `gbar … k` is L♮-convex on `V × ℜ⁻`, for every `k`. -/
theorem theorem_4 (L : ℕ) (hL : 0 < L) (μ : Measure ℝ) [IsProbabilityMeasure μ]
(hμ : μ (Set.Iio 0) = 0) (hμint : Integrable id μ)
(c hh p γ : ℝ) (hc : 0 ≤ c) (hhh : 0 ≤ hh) (hp : 0 ≤ p) (hγ0 : 0 < γ) (hγ1 : γ ≤ 1) (k : ℕ) :
LNatConvexOn (V L) (fbar c hh p γ μ k) ∧
LNatConvexOn (VxNeg L) (liftG (gbar c hh p γ μ k)) := by sorry
end ZipkinLostSales.LNatural
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.