Theorem 2.2 — the β-optimal inspect and revise regions are convex
ProvedRossQC.Regions.theorem_2_2convexitydynamic-programmingp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1quality-control
In the general model of Ross's §2, let the -optimal inspect region be the set of beliefs with , and the -optimal revise region the set of with . Then
The produce region is not claimed to be convex; this asymmetry is what allows an optimal policy with four regions in the two-state model.
Formalization Note The regions are taken inside , since is only meaningful on beliefs.
Preamble
import Mathlib import Definitions.Def_RossQC_Regions_Model
Formal statement
namespace RossQC.Regions
/-- Ross, *Quality Control under Markovian Deterioration*, Management Science 17(9):587–596 (1971),
DOI 10.1287/mnsc.17.9.587, p. 589, Theorem 2.2: "Both the β-optimal inspect and revise regions are
convex."
Under the standing hypotheses of §1–§2 (`Model.Valid`), the β-optimal inspect region and the
β-optimal revise region (Definition, p. 588: the beliefs `P ∈ S` at which inspecting, resp.
revising, attains the minimum in (1)) are convex subsets of `ι → ℝ`.
**Formalization Note.** The regions are subsets of the state space `S` (`Model.region`), as the
paper's `{P : …}` ranges over `S`. `V_β` is the limit of value iteration (`Model.Vβ`). -/
theorem theorem_2_2 {ι : Type*} [Countable ι] (M : Model ι) (hM : M.Valid) :
Convex ℝ M.inspectRegion ∧ Convex ℝ M.reviseRegion := by sorry
end RossQC.Regions
Source
Ross, Quality Control under Markovian Deterioration, Management Science 17(9):587–596 (1971), DOI 10.1287/mnsc.17.9.587, p. 589, Theorem 2.2; Definition of the regions, p. 588
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.