Proposition C.5.2 — an ATAS sending excess probability to a finite set is conforming
ProvedSennottDP.ChainASM.atas_excess_to_finite_set_conformingLet be a standard Markov chain with costs on a denumerable state space , and let be a finite nonempty subset of . Let be an augmentation type approximating sequence whose augmentation distributions send the excess probability to :
Then the ATAS is conforming: for large each is unichain with in its positive recurrent class, and , for all . Moreover, if is contained in the positive recurrent class of , the ATAS is also conforming on : and for .
Truncating a countable chain to finite state spaces and returning the truncated probability to a fixed finite set is the construction the book uses to compute average-cost optimal policies for queueing models; this result guarantees that the truncated chains' first passage moments, steady state probabilities and average costs converge.
Formalization Note is the communicating class of , which for a standard chain is its positive recurrent class (Proposition C.2.6). The augmentation distributions are a function ; the excess condition sums over .
import Mathlib import Definitions.Def_SennottDP_ChainASM_MarkovChain import Definitions.Def_SennottDP_ChainASM_ApproxSeq open scoped ENNReal NNReal open Filter Topology open Classical
namespace SennottDP.ChainASM
theorem atas_excess_to_finite_set_conforming {S : Type*} [Countable S] [Infinite S]
(Γ : MC S) (z : S) (hz : Γ.IsZStandard z) (G : Finset S) (hG : G.Nonempty)
(AS : ApproxSeq Γ) (q : ℕ → S → S → S → ℝ≥0∞) (hq : AS.IsATASWith q)
(hexcess : AS.SendsExcessTo q (↑G : Set S)) :
AS.IsConforming z ∧
((↑G : Set S) ⊆ Γ.commClass z → AS.IsConformingOn (Γ.commClass z)) := by sorry
end SennottDP.ChainASM
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.