Lemma 1.3, part 2 — if a ≥ a′ implies P^a ≥tp P^{a′}, then πP^a ≥r πP^{a′}
ProvedLovejoyPOMDP.Monotone.lemma_1_3_2likelihood-ratio-ordermarkov-chainp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1total-positivity
Consider a finite POMDP as in §2 of the paper, with a completely ordered action set . Suppose that in implies . Then for every belief and all in ,
Larger actions thus push the predicted state distribution up in the likelihood ratio order. This is hypothesis (a) of Lemma 1.2(3) and enters the proof of Lemma 2.3.
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 2.
If `a ≥ a'` in `A` implies `P^a ≥tp P^{a'}`, then `π ∈ Π(S)` and `a ≥ a'` imply
`πP^a ≥r πP^{a'}`. -/
theorem lemma_1_3_2 {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 a' : A, a' ≤ a → TPGE (M.P a) (M.P a')) {π : S → ℝ}
(hπ : π ∈ stdSimplex ℝ S) {a a' : A} (haa' : a' ≤ a) :
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 2
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.