is convex on and as
ProvedInventoryControl.s_policy_cost_convexdiscrete-optimizationinventory-controloperations-researchreorder-point
For a discrete lead-time demand with finite mean and costs , , the -policy cost rate is a convex function of the inventory position , in the sense that its increments are nondecreasing,
and as .
Both follow from the increment identity : the distribution function is nondecreasing, and the increments tend to as and to as . These are the two properties Sect. 6.1.1.1 needs of : convexity makes the best set of consecutive positions grow by one neighbour at a time, and the growth at infinity guarantees that every cost minimization over is attained.
Formalization Note Divergence is stated as Tendsto g (cocompact ℤ) atTop, that is,
along in both directions.
Preamble
import Definitions.Def_InventoryControl_rqDiscrete
Formal statement
namespace InventoryControl
theorem s_policy_cost_convex (D : DiscreteDemand) (h b1 : ℝ) (hh : 0 < h) (hb : 0 < b1) :
(∀ k : ℤ, sPolicyCost D h b1 (k + 1) - sPolicyCost D h b1 k
≤ sPolicyCost D h b1 (k + 2) - sPolicyCost D h b1 (k + 1))
∧ Filter.Tendsto (sPolicyCost D h b1) (Filter.cocompact ℤ) Filter.atTop := by sorry
end InventoryControl
Source
Sven Axsäter, Inventory Control, 3rd ed., Springer 2015, DOI 10.1007/978-3-319-15729-0, p. 108, Sect. 6.1.1.1: 'The results in Sect. 5.9.1 imply that g(k) is a convex function of the inventory position k. Furthermore g(k) -> infinity as |k| -> infinity'; restated p. 114 after Eq. (6.20)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.