Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 13(b) — IVI↓opt,VIVI_{\downarrow opt,V}IVI↓opt,V​ and IVI↑pes,VIVI_{\uparrow pes,V}IVI↑pes,V​ are contraction mappings

Proved
BoundedParamMDP.IntervalEval.theorem13b_contraction

by mikedeng1 · Oct 7, 2026 · Mathlib 0df444a (Lean v4.33.1)

contractiondynamic-programmingmarkov-decision-processesp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1

Let M↕M_\updownarrowM↕​ be a bounded-parameter MDP with finite state set QQQ and finite nonempty action set AAA. For a value function VVV, IVI↓opt,V(V′)=IVI↓opt([V′,V])IVI_{\downarrow opt,V}(V')=IVI_{\downarrow opt}([V',V])IVI↓opt,V​(V′)=IVI↓opt​([V′,V]) and IVI↑pes,V(V′)=IVI↑pes([V,V′])IVI_{\uparrow pes,V}(V')=IVI_{\uparrow pes}([V,V'])IVI↑pes,V​(V′)=IVI↑pes​([V,V′]) are maps from value functions to value functions; their action choices are governed by the sets ρV(p)\rho_V(p)ρV​(p) and σV(p)\sigma_V(p)σV​(p).

Theorem 13(b). For any value function VVV (with its associated action set selection functions ρV\rho_VρV​ and σV\sigma_VσV​), IVI↓opt,VIVI_{\downarrow opt,V}IVI↓opt,V​ and IVI↑pes,VIVI_{\uparrow pes,V}IVI↑pes,V​ are contraction mappings: each TTT of them has some λ∈[0,1)\lambda\in[0,1)λ∈[0,1) with

∥Tv−Tu∥≤λ∥v−u∥for all value functions u,v.\|Tv-Tu\|\le\lambda\|v-u\|\qquad\text{for all value functions }u,v.∥Tv−Tu∥≤λ∥v−u∥for all value functions u,v.

This is what makes the lower bounds of optimistic (upper bounds of pessimistic) interval value iteration converge once the other bound has converged.

Formalization Note. No relation between VVV and the argument V′V'V′ is assumed, so [V′,V][V',V][V′,V] may be an improper interval, as on the page. The modulus may depend on VVV (the paper's statement is per VVV).

Preamble
import Mathlib
import Definitions.Def_BoundedParamMDP_Optimal_MDP
import Definitions.Def_BoundedParamMDP_Optimal_BMDP
import Definitions.Def_BoundedParamMDP_IntervalEval_IntervalPolicyEvaluation
import Definitions.Def_BoundedParamMDP_IntervalEval_IntervalValueIteration
Formal statement
namespace BoundedParamMDP.IntervalEval

/-- Theorem 13(b) (Givan–Leach–Dean 2000, p. 25; proof p. 47): for any value function `V`
(with its action selection sets `ρ_V`, `σ_V`), the maps `IVI↓_{opt,V}` and `IVI↑_{pes,V}`
from value functions to value functions are contraction mappings. -/
theorem theorem13b_contraction {Q A : Type*} [Fintype Q] [Fintype A] [Nonempty A]
    (B : BoundedParamMDP.Optimal.BMDP Q A) (V : Q → ℝ) :
    IsContraction (IVIloOptV B V) ∧ IsContraction (IVIhiPesV B V) := by sorry

end BoundedParamMDP.IntervalEval
Source
Givan, Leach, Dean, Bounded-parameter Markov decision processes, Artificial Intelligence (2000), DOI 10.1016/S0004-3702(00)00047-3, manuscript of May 22, 2000, p. 25, Theorem 13(b) (proof p. 47)
Human review
  • Endorsed by Shuze Chen · Oct 7, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 7, 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