Retailer comparison:
ProvedOWMR.fifo_retailer_comparisonThis is the retailer comparison in Step 1 of the proof of Theorem 1 (Section 5.4 of the source), which refines Proposition 1.
Consider an OWMR network that satisfies the standing assumptions, admissible choices for the FIFO policy of Definition 1, and a retailer . Let and suppose and . Then the augmented cost of the pair that retailer operates satisfies
Here is the long-run average cost of that pair in retailer 's single-location problem with actual setup cost plus its mirror cost, and is the optimal average cost of that problem with setup cost . For , retailer operates the cheaper, in augmented cost, of a canonical optimum at planning setup cost and its best half window; for it operates base stock at a minimizer of .
With the factor is , as in Proposition 1, and it falls toward as grows. This is the retailer part of the upper bound in Theorem 1.
Formalization Note The augmented cost is N.augCost i (c.s i) (c.S i), and the factor enters through ENNReal.ofReal.
import Definitions.Def_owmr_fifo_policy open scoped ENNReal
namespace OWMR
theorem fifo_retailer_comparison {n : ℕ} (N : Network n) (hN : N.Standing)
(m : ℝ) (hm : 1 ≤ m) (c : FIFOChoice n) (hc : IsFIFOChoice N c) (i : Fin n)
(hH : m * N.h ≤ N.H i) (hp : N.h ≤ N.p) :
N.augCost i (c.s i) (c.S i)
≤ ENNReal.ofReal (5 / 4 + 1 / (4 * m)) * N.J i (5 * N.K i / 3) := by sorry
end OWMR