Section 3.4.2, p. 61 — the smallest s with Σ_{x≤s} p(x|µ) ≥ 1/(1+θc) minimizes f(s) = (1+θc)B(s) + θcs
ProvedServiceParts.StockLevels.optimal_stock_criteriondiscrete-convexityinventorylagrangian-relaxationp2o-batch-b23ap2o-gran-per-chapterp2o-plan-bookp2o-v1
Let an item have compound Poisson demand with steady-state probabilities and expected backorders , unit cost , and let be a Lagrange multiplier. Put
Then:
- is discretely convex: for all ;
- ;
- if and only if
- some satisfies this inequality;
- the smallest such , call it , is an optimal stock level: for every , and for every .
This solves the single-item subproblem of the Lagrangian relaxation of Problem 4, which separates by item type.
Formalization Note The book allows in (3.38). Here and are assumed: at the threshold is , which the distribution function of a compound Poisson law never reaches, and has no minimizer. Optimality is against every nonnegative integer .
Preamble
import Mathlib import Definitions.Def_ServiceParts_StockLevels_Basic import Definitions.Def_ServiceParts_StockLevels_CompoundPoissonDemand import Definitions.Def_ServiceParts_StockLevels_Problems
Formal statement
namespace ServiceParts.StockLevels
theorem optimal_stock_criterion (d : CompoundPoissonDemand) (c θ : ℝ) (hc : 0 < c)
(hθ : 0 < θ) :
(∀ s, 0 ≤ fdiff2 (relaxedCost d c θ) s) ∧
(∀ s, fdiff (relaxedCost d c θ) s = -(1 + θ * c) * (1 - d.readyRate s) + θ * c) ∧
(∀ s, 0 ≤ fdiff (relaxedCost d c θ) s ↔ s ∈ stockSet d c θ) ∧
(stockSet d c θ).Nonempty ∧
∀ sStar : ℕ, IsLeast (stockSet d c θ) sStar →
(∀ s, relaxedCost d c θ sStar ≤ relaxedCost d c θ s) ∧
(∀ s, s < sStar → relaxedCost d c θ sStar < relaxedCost d c θ s) := by sorry
end ServiceParts.StockLevels
Source
Muckstadt, Analysis and Algorithms for Service Parts Supply Chains, Springer 2005, DOI 10.1007/b138879, p. 61, Section 3.4.2 (f(s), Δf(s), and the optimal stock level s*)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.