Each of the Clark-Scarf recursion is convex under sequential minimization
ProvedSupplyChainTheory.cs_g_convexconvexitymultiechelonoperations-researchsupply-chain
For echelon holding costs , stockout cost , lead-time demands of finite mean, and a base-stock vector built by the sequential minimization (6.26) (each minimizes ), every function , , is convex on .
This is the remark the book attaches to Theorem 6.3: at each iteration one minimizes a single-variable convex function. The induction is that is convex, adds a linear term to , is an expectation of translates of , and is convex because is a minimizer of the convex : it is to the left of and constant to the right. The last step is where sequential optimality is used; for an arbitrary the functions need not be convex.
Preamble
import Definitions.Def_SupplyChainTheory_multiechelon
Formal statement
namespace SupplyChainTheory
theorem cs_g_convex (N : ℕ) (h : ℕ → ℝ) (p : ℝ) (D : ℕ → MeasureTheory.Measure ℝ) (S : ℕ → ℝ)
[∀ j, MeasureTheory.IsProbabilityMeasure (D j)]
(hD : ∀ j, MeasureTheory.Integrable (fun x => x) (D j))
(hh : ∀ j, 0 ≤ h j) (hp : 0 ≤ p) (hS : CSSequential N h p D S) (j : ℕ) (hj : j ≤ N) :
ConvexOn ℝ Set.univ (csG N h p D S j) := by sorry
end SupplyChainTheory
Source
Lawrence V. Snyder and Zuo-Jun Max Shen, Fundamentals of Supply Chain Theory, 2nd ed., Wiley 2019, DOI 10.1002/9781119584445, p. 196, Sect. 6.2.2, the remark after Theorem 6.3: 'Moreover, gj(y) is known to be convex, so at each iteration we only need to minimize a single-variable, convex function'
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.