Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proof of Theorem 5.9 — every policy has average cost at least ρ* = lim (1 − β_n) J*_{β_n}(i)

Proved
ArapostathisAC.SennottACOI.avgCost_ge_limit

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

average-costp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1tauberianvanishing-discount

In the countable-state controlled Markov process of §5, assume Assumption 5.14 (Jβ∗<∞J^*_\beta<\inftyJβ∗​<∞ for all β∈(0,1)\beta\in(0,1)β∈(0,1)). Let (βn)(\beta_n)(βn​) be a sequence in (0,1)(0,1)(0,1) with βn→1\beta_n\to1βn​→1, let i∈Si\in Si∈S, and suppose that

lim⁡n→∞(1−βn) Jβn∗(i)=ρ∗.\lim_{n\to\infty}(1-\beta_n)\,J^*_{\beta_n}(i)=\rho^*.n→∞lim​(1−βn​)Jβn​∗​(i)=ρ∗.

Then every admissible policy π∈Π\pi\in\Piπ∈Π (history dependent, possibly randomized) satisfies

J(i,π)=lim sup⁡N→∞1N Eiπ∑t=0N−1c(Xt,At) ≥ ρ∗.J(i,\pi)=\limsup_{N\to\infty}\frac1N\,E^\pi_i\sum_{t=0}^{N-1}c(X_t,A_t)\ \ge\ \rho^*.J(i,π)=N→∞limsup​N1​Eiπ​t=0∑N−1​c(Xt​,At​) ≥ ρ∗.

This is the lower bound of the proof of Theorem 5.9, obtained there from the Tauberian Theorem A.2; combined with J(i,f)≤ρ∗J(i,f)\le\rho^*J(i,f)≤ρ∗ it makes the limit policy fff AC-optimal.

Formalization Note The paper's ρ∗\rho^*ρ∗ is the limit from the first paragraph of the proof; the statement holds for the limit along any sequence βn→1\beta_n\to1βn​→1 along which it exists. J(i,π)J(i,\pi)J(i,π) lies in [0,∞][0,\infty][0,∞] and the bound reads ENNReal.ofReal ρ ≤ avgCost M π i; ρ∗≥0\rho^*\ge0ρ∗≥0 automatically.

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]

/-- Proof of Theorem 5.9 (p. 308): "By Theorem A.2 in the Appendix, J(i, π) ≥ ρ* for any π ∈ Π."
Under Assumption 5.14, if `β_n ∈ (0, 1)`, `β_n → 1` and `(1 − β_n) J*_{β_n}(i) → ρ*` for a state
`i`, then every admissible policy `π` (history dependent, randomized) has average cost
`J(i, π) ≥ ρ*`. -/
theorem avgCost_ge_limit (M : ArapostathisAC.VanishingDiscount.CMP A) (h14 : Assumption5_14 M)
    (β : ℕ → ℝ) (hβ : ∀ n, β n ∈ Set.Ioo (0 : ℝ) 1) (hβ1 : Tendsto β atTop (𝓝 1))
    (ρ : ℝ) (i : ℕ)
    (hρ : Tendsto (fun n => (1 - β n) * (ArapostathisAC.VanishingDiscount.discValue M (β n) i).toReal) atTop (𝓝 ρ))
    (π : ArapostathisAC.VanishingDiscount.Policy M) :
    ENNReal.ofReal ρ ≤ ArapostathisAC.VanishingDiscount.avgCost M π i := 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 ("By Theorem A.2 in the Appendix, J(i, π) ≥ ρ* for any π ∈ Π")
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