iff , and the minimum increases with
ProvedInventoryControl.rq_discrete_marginal_monoTwo facts about the marginal cost of the -th unit of the batch. Under a discrete lead-time demand with finite mean, ordering cost , demand rate , and costs , , let and be optimal reorder points for the batch quantities and , and write for the cost of the cheaper neighbour of the optimal window. Then
- if and only if ;
- .
The first is read off Eq. (6.7): is a weighted average of and with weights and , so it exceeds exactly when does. The second is the book's "obvious" step: as the optimal window grows, the neighbours it has not yet absorbed lie further out on a convex and can only cost more. Together they are what makes the stopping rule of Sect. 6.1.1.1 correct.
Formalization Note The reorder points may be any optimal ones for their batch quantities; by Eq. (6.7) the value does not depend on the choice.
import Definitions.Def_InventoryControl_rqDiscrete
namespace InventoryControl
theorem rq_discrete_marginal_mono (D : DiscreteDemand) (h b1 A μ : ℝ) (hh : 0 < h) (hb : 0 < b1)
(Q : ℕ) (hQ : 0 < Q) (Rstar Rnext : ℤ)
(hR : ∀ R : ℤ, rqDiscreteCost D h b1 A μ Rstar Q ≤ rqDiscreteCost D h b1 A μ R Q)
(hR' : ∀ R : ℤ, rqDiscreteCost D h b1 A μ Rnext (Q + 1) ≤ rqDiscreteCost D h b1 A μ R (Q + 1)) :
(rqDiscreteCost D h b1 A μ Rstar Q ≤ rqDiscreteCost D h b1 A μ Rnext (Q + 1)
↔ rqDiscreteCost D h b1 A μ Rstar Q
≤ min (sPolicyCost D h b1 Rstar) (sPolicyCost D h b1 (Rstar + Q + 1)))
∧ min (sPolicyCost D h b1 Rstar) (sPolicyCost D h b1 (Rstar + Q + 1))
≤ min (sPolicyCost D h b1 Rnext) (sPolicyCost D h b1 (Rnext + (Q + 1) + 1)) := by sorry
end InventoryControl
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.