Section 5 — if every is indecomposable, is a strict vectorial maximum of
ProvedDEpenoux.LinearProgram.strict_vectorial_maximumIn d'Epenoux's production–inventory model with stock levels , a stochastic kernel , real costs and discount , assume that for every strategy (every map with ) the transition matrix is indecomposable: for all there is with . Let be a solution of (7),
Then is a strict vectorial maximum of the constraint set of (9): for every with ,
This is the sufficient condition stated at the end of the strict-maximum paragraph of Section 5.
Formalization Note Only the sufficient condition ("it is sufficient that they form an indecomposable group with respect to all the transition matrices ") is formalized; the paper's necessary and sufficient condition in terms of is not, because it comes with no argument and its meaning when several strategies are optimal is unclear. "Indecomposable group" is read as irreducibility of the matrix, as in (5).
import Mathlib import Definitions.Def_DEpenoux_LinearProgram_Model
namespace DEpenoux.LinearProgram
theorem strict_vectorial_maximum (σ : ℕ) (p d : Fin (σ + 1) → Fin (σ + 1) → ℝ)
(hp0 : ∀ j s, 0 ≤ p j s) (hp1 : ∀ j, ∑ s, p j s = 1)
(lam : ℝ) (hlam0 : 0 < lam) (hlam1 : lam < 1)
(hirr : ∀ J : Fin (σ + 1) → Fin (σ + 1), IsStrategy J →
∀ i k, ∃ n : ℕ, 0 < (stratMatrix p J ^ n) i k)
(uStar : Fin (σ + 1) → ℝ)
(huStar : BertsekasDiscountedBellmanOp (model σ p d hp0 hp1) lam uStar = uStar)
(u : Fin (σ + 1) → ℝ) (hu : u ∈ setA p d lam) (hne : u ≠ uStar) :
∀ i, u i < uStar i := by sorry
end DEpenoux.LinearProgram
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.