Theorem D.1 (DFR Demand): , ,
ProvedOWMR.dfr_guaranteeLet and . Consider an OWMR network with retailers under the standing assumptions of Section 3: demand vectors are i.i.d. over periods and may be dependent across retailers, and the lead times are arbitrary finite integers. Suppose , and for every retailer, and that each retailer with has an absolutely continuous DFR distribution conditional on positive demand (DFRPositive). Let , and let be any choices satisfying Definition D.1 (IsDFRChoice N θ c). Then, from every valid initial state , the policy , the FIFO policy fifoPolicy N x c, is nonanticipative and feasible, and
where is the long-run average cost (2) and is the infimum of over all nonanticipative, feasible, measurable policies, which may randomize independently of demand.
import Definitions.Def_owmr_dfr_policy open scoped ENNReal
namespace OWMR
theorem dfr_guarantee {n : ℕ} (hn : 0 < n) (N : Network n) (hN : N.Standing)
(hH : ∀ i, N.h ≤ N.H i) (hp : N.h ≤ N.p) (hK : ∀ i, N.K i ≤ (dfrGamma - 1) * N.K0)
(hD : ∀ i, 0 < N.K i → N.DFRPositive i)
(x : InitState n) (hx : x.Valid) (c : FIFOChoice n) (hc : IsDFRChoice N (1 - 1 / dfrGamma) c) :
(fifoPolicy N x c).toPolicy.Nonanticipative ∧ (fifoPolicy N x c).toPolicy.Feasible N x ∧
(fifoPolicy N x c).avgCost N x ≤ ENNReal.ofReal dfrGamma * OPT N x := by sorry
end OWMRConfirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.