Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Remark 5.8(a), the half used — an ACOI with h bounded below gives J(i, f) ≤ ρ

Proved
ArapostathisAC.SennottACOI.acoi_avgCost_le

by mikedeng1 · Oct 5, 2026 · Mathlib 0df444a (Lean v4.33.1)

average-costaverage-cost-optimality-inequalityp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1verification-theorem

In the countable-state controlled Markov process of §5, let ρ∈R\rho\in\mathbb Rρ∈R, let h:S→Rh:S\to\mathbb Rh:S→R be bounded below, and let f∈ΠSDf\in\Pi_{SD}f∈ΠSD​. Suppose that for every state iii the series ∑jP(j∣i,f(i))h(j)\sum_jP(j\mid i,f(i))h(j)∑j​P(j∣i,f(i))h(j) converges and

ρ+h(i) ≥ c(i,f(i))+∑jP(j∣i,f(i)) h(j).\rho+h(i)\ \ge\ c\big(i,f(i)\big)+\sum_jP\big(j\mid i,f(i)\big)\,h(j).ρ+h(i) ≥ c(i,f(i))+j∑​P(j∣i,f(i))h(j).

Then the average cost of fff satisfies J(i,f)≤ρJ(i,f)\le\rhoJ(i,f)≤ρ for every state iii.

In the proof of Theorem 5.9 this gives J(i,f)≤ρ∗J(i,f)\le\rho^*J(i,f)≤ρ∗ for the limit policy fff ("Since h(⋅)h(\cdot)h(⋅) is bounded below, the proof of Theorem 5.1 can be modified to show that J(i,f)≤ρ∗J(i,f)\le\rho^*J(i,f)≤ρ∗").

Formalization Note The paper's Remark 5.8(a) states more: that under (5.16) with hhh bounded below, ρ\rhoρ is the optimal average cost and every minimizing fff is AC-optimal. That does not follow for an arbitrary scalar ρ\rhoρ: with h≡0h\equiv0h≡0 and ρ=sup⁡c\rho=\sup cρ=supc (bounded ccc), (5.16) holds, yet ρ\rhoρ need not be the optimal cost. Only the inequality J(i,f)≤ρJ(i,f)\le\rhoJ(i,f)≤ρ holds and is used, so only it is stated. The average cost is a lim sup⁡\limsuplimsup in [0,∞][0,\infty][0,∞] and ρ\rhoρ enters as ENNReal.ofReal ρ.

Preamble
import Mathlib
import Definitions.Def_ArapostathisAC_SennottACOI_CMP
import Definitions.Def_ArapostathisAC_SennottACOI_Assumptions

open MeasureTheory ProbabilityTheory Filter Topology
open scoped ENNReal
Formal statement
namespace ArapostathisAC.SennottACOI

variable {A : Type*} [MetricSpace A] [MeasurableSpace A] [BorelSpace A]

/-- Remark 5.8(a) (p. 308), the half used in the proof of Theorem 5.9 ("Since h(·) is bounded
below, the proof of Theorem 5.1 can be modified to show that J(i, f) ≤ ρ*"): if `ρ` is a real
number, `h : S → ℝ` is bounded below, `f ∈ Π_SD`, and for every state `i` the series
`Σ_j P(j | i, f(i)) h(j)` converges with `ρ + h(i) ≥ c(i, f(i)) + Σ_j P(j | i, f(i)) h(j)`, then
`J(i, f) ≤ ρ` for every state `i`. The printed remark further claims that `ρ` is the optimal
average cost, which does not follow; that part is not stated. -/
theorem acoi_avgCost_le (M : ArapostathisAC.VanishingDiscount.CMP A) (ρ : ℝ) (h : ℕ → ℝ) (hb : BddBelow (Set.range h))
    (f : StationaryPolicy M) (hsum : ∀ i, Summable (fun j => ArapostathisAC.VanishingDiscount.prob M i (f.1 i) j * h j))
    (hineq : ∀ i, M.c i (f.1 i) + ∑' j, ArapostathisAC.VanishingDiscount.prob M i (f.1 i) j * h j ≤ ρ + h i) (i : ℕ) :
    ArapostathisAC.VanishingDiscount.avgCost M f.toPolicy i ≤ ENNReal.ofReal ρ := by sorry

end ArapostathisAC.SennottACOI
Source
Arapostathis, Borkar, Fernández-Gaucherand, Ghosh, Marcus, Discrete-time controlled Markov processes with average cost criterion: a survey, SIAM J. Control Optim. 31(2) (1993), p. 308, proof of Theorem 5.9 ("Since h(·) is bounded below, …") and Remark 5.8(a), (5.16)
Human review
  • Endorsed by Shuze Chen · Oct 5, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 5, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me