Lemma 1.3, part 1 — a TP₂ transition matrix preserves the MLR order: π ≥r π′ ⇒ πP^a ≥r π′P^a
ProvedLovejoyPOMDP.Monotone.lemma_1_3_1likelihood-ratio-ordermarkov-chainp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1total-positivity
Consider a finite POMDP as in §2 of the paper. Suppose that for each action the transition matrix is on . Then for every and all beliefs ,
Together with Lemma 1.2(2), this shows that the Bayes update is MLR-nondecreasing in the prior .
Preamble
import Mathlib import Definitions.Def_LovejoyPOMDP_Monotone_Model
Formal statement
namespace LovejoyPOMDP.Monotone
/-- Lovejoy, *Some Monotonicity Results for Partially Observed Markov Decision Processes*, Oper. Res. 35(5):736–743 (1987), DOI 10.1287/opre.35.5.736, p. 739, Lemma 1.3, part 1.
If every `P^a` is TP₂ on `S × S`, then for every `a ∈ A` and `π ≥r π'` in `Π(S)`,
`πP^a ≥r π'P^a`. -/
theorem lemma_1_3_1 {S O A : Type*} [Fintype S] [Fintype O] [Fintype A]
[LinearOrder S] [LinearOrder O] [LinearOrder A] [Nonempty S] [Nonempty O] [Nonempty A]
(M : POMDP S O A)
(hP : ∀ a, TP2 (M.P a)) (a : A) {π π' : S → ℝ}
(hπ : π ∈ stdSimplex ℝ S) (hπ' : π' ∈ stdSimplex ℝ S) (h : MLRGE π π') :
MLRGE (M.predict π a) (M.predict π' a) := by sorry
end LovejoyPOMDP.Monotone
Source
Lovejoy, Some Monotonicity Results for Partially Observed Markov Decision Processes, Oper. Res. 35(5):736–743 (1987), DOI 10.1287/opre.35.5.736, p. 739, Lemma 1.3, part 1
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.