Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 4.2.2 — the AMD-value-vector is the smallest AMD-superharmonic vector

Proved
KallenbergLP.AverageLP.theorem_4_2_2

by mikedeng1 · Oct 6, 2026 · Mathlib 0df444a (Lean v4.33.1)

average-rewarddynamic-programmingmarkov-decision-processp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-bookp2o-v1superharmonic

In a finite Markov decision model with stochastic transition rows, let ϕi=sup⁡Rϕi(R)\phi_i=\sup_R\phi_i(R)ϕi​=supR​ϕi​(R) be the AMD-value-vector, the supremum over all policies of the lim inf average reward. A vector ϕ~\tilde\phiϕ~​ is AMD-superharmonic if there is a vector u~\tilde uu~ with

ϕ~i≥∑jpiajϕ~jandϕ~i+u~i≥ria+∑jpiaju~jfor all a∈A(i), i∈E.\tilde\phi_i\ge\sum_jp_{iaj}\tilde\phi_j\quad\text{and}\quad\tilde\phi_i+\tilde u_i\ge r_{ia}+\sum_jp_{iaj}\tilde u_j\qquad\text{for all }a\in A(i),\ i\in E.ϕ~​i​≥j∑​piaj​ϕ~​j​andϕ~​i​+u~i​≥ria​+j∑​piaj​u~j​for all a∈A(i), i∈E.

Then ϕ\phiϕ is AMD-superharmonic, and ϕ≤ϕ~\phi\le\tilde\phiϕ≤ϕ~​ componentwise for every AMD-superharmonic vector ϕ~\tilde\phiϕ~​.

This is what makes the linear program (4.2.10), which minimises ∑jβjϕ~j\sum_j\beta_j\tilde\phi_j∑j​βj​ϕ~​j​ over superharmonic pairs, have ϕ\phiϕ as the first component of its optimal solutions.

Preamble
import Mathlib
import Definitions.Def_MarkovDecisionProcesses_AverageReward
import Definitions.Def_BlackwellDiscreteDP_NearOne_LimitMatrix
import Definitions.Def_KallenbergLP_AverageLP_Model
open MarkovDecisionProcesses BlackwellDiscreteDP.NearOne Matrix Filter
Formal statement
namespace KallenbergLP.AverageLP

variable {S A : Type*} [Fintype S] [Fintype A] [DecidableEq S] [DecidableEq A]

/-- **Theorem 4.2.2.** The AMD-value-vector `φ` is the smallest AMD-superharmonic vector: `φ` is
AMD-superharmonic, and `φ ≤ φ̃` componentwise for every AMD-superharmonic `φ̃`.

Kallenberg, Linear Programming and Finite Markovian Control Problems, Mathematical Centre Tracts 148, Mathematisch Centrum, Amsterdam 1983, p. 99, Theorem 4.2.2.

**Formalization Note.** `φ_i = sup_R φ_i(R)` (`optGainInf`) is the supremum over all
history-dependent randomized policies of the lim inf average reward. -/
theorem theorem_4_2_2 (M : StationaryMDP S A) :
    IsAMDSuperharmonic M (optGainInf M) ∧
      ∀ φ' : S → ℝ, IsAMDSuperharmonic M φ' → ∀ i, optGainInf M i ≤ φ' i := by sorry

end KallenbergLP.AverageLP
Source
Kallenberg, Linear Programming and Finite Markovian Control Problems, Mathematical Centre Tracts 148, Mathematisch Centrum, Amsterdam 1983, p. 99, Theorem 4.2.2; Definition 4.2.1, pp. 98–99
Human review
  • Endorsed by Shuze Chen · Oct 6, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 6, 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