Observation 2.2, p. 8 — N(R*, Δ) ≤ N(T*, Δ) for every Δ ≥ 0
ProvedContJRP.Align.observation_2_2_rep_ordersapproximation-schemejoint-replenishmentp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Under the standing assumptions below, the representatives place no more joint orders than the policy they are drawn from: for every ,
Standing assumptions. Throughout, ; is a policy with and all , ; satisfies (2), ; and is any choice of representatives, being some in the active segment .
This links the joint orders of the representatives to those of in the density bound (3).
Formalization Note Segments are 0-indexed in Lean (the paper's is Lean index , so is R 0). The statement is proved for every positive reference policy , not only an optimal one, and does not assume the difficult regime : the page's argument uses only (2).
Preamble
import Mathlib import Definitions.Def_ContJRP_Align_Model import Definitions.Def_ContJRP_Align_Alignment import Definitions.Def_ContJRP_Align_Policy
Formal statement
namespace ContJRP.Align
theorem observation_2_2_rep_orders (ε : ℝ) (hε0 : 0 < ε) (hε1 : ε < 1 / 2)
{n : ℕ} [NeZero n] (Tstar : Fin n → ℝ) (hTstar : ∀ i, 0 < Tstar i)
(Tt : ℝ) (hTt1 : (1 - ε / 2) * tmin Tstar ≤ Tt) (hTt2 : Tt ≤ tmin Tstar)
(Rstar : ℕ → ℝ) (hR : IsRepChoice ε Tt Tstar Rstar) :
∀ Δ : ℝ, 0 ≤ Δ → numOrders (active ε Tt Tstar) Rstar Δ ≤ numOrders Finset.univ Tstar Δ := by sorry
end ContJRP.Align
Source
Segev, The Continuous-Time Joint Replenishment Problem: ϵ-Optimal Policies via Pairwise Alignment, arXiv:2302.09941v1, p. 8, Observation 2.2
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.