Section 3.4.2, pp. 58-61 — the Lagrangian stock levels s_i*(θ) solve Problem 4 with budget b = C(θ)
ProvedServiceParts.StockLevels.lagrangian_stock_levels_optimalConsider item types managed at a single location by policies. Item has stationary compound Poisson demand with steady-state resupply probabilities , expected backorders , mean lead-time demand , and unit cost . Problem 4 (3.40) with budget is
Fix a multiplier . For every item there is a stock level with
and if denotes the smallest one, then the vector is an optimal solution of Problem 4 with budget
every vector of nonnegative integer stock levels with satisfies
Varying therefore traces optimal backorder–investment trade-offs, which is how the book's bisection algorithm for Problem 4 is justified.
Formalization Note The book's text asserts this by combining Everett's theorem (Theorem 10, p. 57, and the remark on p. 58 that varying yields optimal solutions of Problem 3) with the item-wise criterion of p. 61. and are assumed; at the criterion has no solution. The index set of items is any finite type.
import Mathlib import Definitions.Def_ServiceParts_StockLevels_CompoundPoissonDemand import Definitions.Def_ServiceParts_StockLevels_Problems
namespace ServiceParts.StockLevels
theorem lagrangian_stock_levels_optimal {ι : Type*} [Fintype ι]
(d : ι → CompoundPoissonDemand) (c : ι → ℝ) (hc : ∀ i, 0 < c i) (θ : ℝ) (hθ : 0 < θ) :
(∀ i, (stockSet (d i) (c i) θ).Nonempty) ∧
∀ sStar : ι → ℕ, (∀ i, IsLeast (stockSet (d i) (c i) θ) (sStar i)) →
∀ s : ι → ℕ, totalInvestment d c s ≤ totalInvestment d c sStar →
totalBackorders d sStar ≤ totalBackorders d s := by sorry
end ServiceParts.StockLevels
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.