Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 3, p. 1130 — U^k u ≥ (I + A^k B)v for u ≥ v; (I + A^k B)u ≥ U^0 v if also Bu ≥ 0

Proved
ModPolicyIter.Conv.lemma_3

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

markov-decision-processmodified-policy-iterationmonotonicityp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1

Throughout, SSS is a set of states with a σ\sigmaσ-algebra; each policy P∈PP\in\mathcal PP∈P is a Markov transition kernel on SSS with a measurable one-period reward cPc_PcP​, sup⁡Psup⁡s∣cP(s)∣≤M\sup_P\sup_s|c_P(s)|\le MsupP​sups​∣cP​(s)∣≤M; 0≤λ<10\le\lambda<10≤λ<1; VVV is the space of bounded measurable functions with ∥v∥=sup⁡s∣v(s)∣\|v\|=\sup_s|v(s)|∥v∥=sups​∣v(s)∣; and the maximum in Bv=max⁡P{cP+(λP−I)v}Bv=\max_P\{c_P+(\lambda P-I)v\}Bv=maxP​{cP​+(λP−I)v} is attained for every v∈Vv\in Vv∈V, a maximizer being written PvP_vPv​.

Write (I+AkB)x=x+APxkBx(I+A^kB)x=x+A^k_{P_x}Bx(I+AkB)x=x+APx​k​Bx, with APk=∑i=0k(λP)iA^k_P=\sum_{i=0}^k(\lambda P)^iAPk​=∑i=0k​(λP)i and PxP_xPx​ a maximizer at xxx. Let k≥0k\ge 0k≥0 and u,v∈Vu,v\in Vu,v∈V with v≤uv\le uv≤u pointwise. Then

  1. for every maximizer PvP_vPv​ at vvv,
Uku ≥ v+APvkBv;U^ku\ \ge\ v+A^k_{P_v}Bv;Uku ≥ v+APv​k​Bv;
  1. if moreover Bu≥0Bu\ge 0Bu≥0, then for every maximizer PuP_uPu​ at uuu,
u+APukBu ≥ U0v.u+A^k_{P_u}Bu\ \ge\ U^0v.u+APu​k​Bu ≥ U0v.

So one step of modified policy iteration of order kkk is sandwiched between one step of value iteration and one step of UkU^kUk-iteration; this is the comparison behind Theorem 1 and Corollary 1.

Preamble
import Mathlib
import Definitions.Def_ModPolicyIter_Conv_Setting

open MeasureTheory ProbabilityTheory
Formal statement
namespace ModPolicyIter.Conv

/-- Lemma 3 (Puterman–Shin 1978, p. 1130): for `u ≥ v` in `V`, `U^k u ≥ (I + A^k B)v`; if in
addition `u ∈ V_B`, then `(I + A^k B)u ≥ U^0 v`. Here `(I + A^k B)x = x + A^k_x Bx` with `A^k_x`
the truncated series of a maximizer at `x`. -/
theorem lemma_3 {S : Type*} [MeasurableSpace S] {ι : Type*} (m : Model S ι)
    (hmax : MaxAttained m) (k : ℕ) :
    ∀ u v : S → ℝ, IsBM u → IsBM v → (∀ s, v s ≤ u s) →
      (∀ i, Attains m v i → ∀ s, v s + Apow m i k (B m v) s ≤ U m k u s) ∧
      ((∀ s, 0 ≤ B m u s) → ∀ i, Attains m u i →
        ∀ s, U m 0 v s ≤ u s + Apow m i k (B m u) s) := by sorry

end ModPolicyIter.Conv
Source
Puterman and Shin, Modified policy iteration algorithms for discounted Markov decision problems, Management Science 24 (1978), DOI 10.1287/mnsc.24.11.1127, p. 1130, Lemma 3
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).

  • Endorsed by Community (Bot) · Oct 29, 2026

    Confirmed by the moderator at approval.

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