Claim (14) — (h/T) E[Σ_{t=K̃+1}^T Z_t(m)] ≤ 2hm E[|J₁(ρ′)|²]/√T
ProvedAIMInventory.UnknownRho.claim_14Let be i.i.d. nonnegative demands with finite positive mean, fix the non-perishable AIM parameters and put . Let be the queue , , let and (possibly ), and let be the length of the first busy period of the fixed-inflow queue , , i.e. .
If , then for every
Once the queue has emptied after , it is dominated by the stationary-increment queue with inflow , and the claim bounds the cost of the remaining periods by the second moment of a busy period. Together with the bound on it gives Theorem 8.
Formalization Note. The paper writes with the random index outside the expectation; as in (13) and Appendix E, the Lean statement reads it as . The sum is empty when or . Both sides are in , so an infinite makes the bound trivial rather than junk. The hypothesis is implicit in the paper.
import Mathlib import Definitions.Def_AIMInventory_UnknownRho_Model noncomputable section open MeasureTheory open scoped ENNReal NNReal BigOperators
namespace AIMInventory.UnknownRho
/-- Claim (14), p. 14, with the random lower summation limit inside the
expectation, as required by (13) and Appendix E. -/
theorem claim_14 (P : DemandLaw) (c : Parameters) (ρ' : ℝ) (T : ℕ)
(hT : 1 ≤ T) (hρ : 0 < ρ') (hsmall : ρ' < m c)
(hmean : ρ' < CappedBaseStock.mean P) :
ENNReal.ofReal (c.h / (T : ℝ)) *
(∫⁻ d, ∑ t ∈ Finset.Icc 1 T,
if restartTime c ρ' d < (t : ℝ≥0∞) then
ENNReal.ofReal (Z d (m c) t) else 0
∂CappedBaseStock.demandPathLaw P) ≤
ENNReal.ofReal (2 * c.h * m c / Real.sqrt (T : ℝ)) *
busySecondMoment P ρ' := by sorry
end AIMInventory.UnknownRho
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.