Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 2.3 (The Ideal Penalty)

Proved
InfoRelax.IdealPenalty.theorem_2_3_ideal_penalty

by mikedeng1 · Oct 8, 2026 · Mathlib 0df444a (Lean v4.33.1)

dualitydynamic-programminginformation-relaxationp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1

Consider a dynamic program in recursive form: a natural filtration F\mathbb FF with F0={∅,Ω}\mathcal F_0=\{\emptyset,\Omega\}F0​={∅,Ω}, a countable action set, nonempty feasible-action sets At(a0,…,at−1)A_t(a_0,\dots,a_{t-1})At​(a0​,…,at−1​), and bounded period rewards rt(a0,…,at)r_t(a_0,\dots,a_t)rt​(a0​,…,at​) that are Ft\mathcal F_tFt​-measurable, with total reward r=∑trtr=\sum_t r_tr=∑t​rt​ and value functions VtV_tVt​ given by (2). Let G\mathbb GG be a relaxation of F\mathbb FF and let z⋆z^\starz⋆ be the penalty of Proposition 2.2 for the generating functions wt(a)=Vt+1(a0,…,at)w_t(a)=V_{t+1}(a_0,\dots,a_t)wt​(a)=Vt+1​(a0​,…,at​):

z⋆(a)=∑t=0T(E[Vt+1(a)∣Gt]−E[Vt+1(a)∣Ft]).z^\star(a)=\sum_{t=0}^T\Big(\mathbb E[V_{t+1}(a)\mid\mathcal G_t]-\mathbb E[V_{t+1}(a)\mid\mathcal F_t]\Big).z⋆(a)=t=0∑T​(E[Vt+1​(a)∣Gt​]−E[Vt+1​(a)∣Ft​]).

Then:

  1. z⋆z^\starz⋆ is dual feasible, z⋆∈ZFz^\star\in\mathcal Z_{\mathbb F}z⋆∈ZF​;
  2. z⋆z^\starz⋆ is optimal:
sup⁡αF∈AFE[r(αF)]=sup⁡αG∈AGE[r(αG)−z⋆(αG)];(11)\sup_{\alpha_F\in\mathcal A_{\mathbb F}}\mathbb E[r(\alpha_F)]=\sup_{\alpha_G\in\mathcal A_{\mathbb G}}\mathbb E[r(\alpha_G)-z^\star(\alpha_G)];\qquad(11)αF​∈AF​sup​E[r(αF​)]=αG​∈AG​sup​E[r(αG​)−z⋆(αG​)];(11)
  1. if αF∗∈AF\alpha^*_F\in\mathcal A_{\mathbb F}αF∗​∈AF​ attains the supremum on the left of (11), then αF∗\alpha^*_FαF∗​ also attains the supremum on the right;
  2. if G\mathbb GG is the perfect-information relaxation (Gt=F\mathcal G_t=\mathcal FGt​=F for all ttt) and αF∗∈AF\alpha^*_F\in\mathcal A_{\mathbb F}αF∗​∈AF​ is an optimal policy, then r(αF∗)−z⋆(αF∗)=E[r(αF∗)]r(\alpha^*_F)-z^\star(\alpha^*_F)=\mathbb E[r(\alpha^*_F)]r(αF∗​)−z⋆(αF∗​)=E[r(αF∗​)] almost surely.

The ideal penalty closes the duality gap for every information relaxation at once; in the perfect-information case it even removes all randomness from the penalized reward of an optimal policy. It is the target that practical penalties built from approximate value functions try to approximate.

Formalization Note z⋆z^\starz⋆ and the value functions are definitions computed from the data of the DP, not variables constrained by hypotheses. Values are extended reals. The countable action set and bounded rewards are pinned regularity assumptions.

Preamble
import Mathlib
import Definitions.Def_InfoRelax_IdealPenalty_Framework
import Definitions.Def_InfoRelax_IdealPenalty_RecursiveModel
import Definitions.Def_InfoRelax_IdealPenalty_Penalty

open MeasureTheory
Formal statement
namespace InfoRelax.IdealPenalty

/-- **Theorem 2.3 (The Ideal Penalty)**, Brown, Smith & Sun (2010), Oper. Res., DOI
10.1287/opre.1090.0796, p. 5, eq. (11).

In the recursive setting, let `𝔾` be a relaxation of `𝔽` and let `z⋆` be the penalty of
Proposition 2.2 for `w_t(a) = V_{t+1}(a_0, …, a_t)`. Then
1. `z⋆` is dual feasible;
2. `z⋆` is optimal: `sup_{α_F ∈ 𝒜_𝔽} 𝔼[r(α_F)] = sup_{α_G ∈ 𝒜_𝔾} 𝔼[r(α_G) − z⋆(α_G)]` (11);
3. if `α*_F ∈ 𝒜_𝔽` attains the primal supremum, it also attains the supremum on the right of (11);
4. if `𝔾` is the perfect-information relaxation and `α*_F ∈ 𝒜_𝔽` is optimal, then
   `r(α*_F) − z⋆(α*_F) = 𝔼[r(α*_F)]` almost surely.

**Formalization Note.** `z⋆` is the def `idealPenalty`, built from the def `valueFn` (the
recursion (2)); nothing is a free variable. Values in `EReal`. Recursive setting
(`RecursiveDP.Valid`: `𝓕_0 = {∅, Ω}`, nonempty feasible sets depending on past actions,
`𝓕_t`-measurable bounded period rewards depending on `a_0, …, a_t`); pinned countable action type
with the discrete σ-algebra. "Perfect information" is `𝔾 = perfectInfo` (`𝒢_t = 𝓕` for all `t`). -/
theorem theorem_2_3_ideal_penalty {Ω X : Type*} [mΩ : MeasurableSpace Ω] [MeasurableSpace X]
    [DiscreteMeasurableSpace X] [Countable X] [DecidableEq X] {T : ℕ}
    (μ : Measure Ω) [IsProbabilityMeasure μ] (M : RecursiveDP Ω X T) (hM : M.Valid)
    (𝔾 : Filtration (Fin (T + 1)) mΩ) (h𝔾 : IsRelaxation M.𝔽 𝔾) :
    idealPenalty M μ 𝔾 ∈ dualFeasible μ M.A M.𝔽 ∧
      primalValue μ M.A M.𝔽 M.r = dualBound μ M.A 𝔾 M.r (idealPenalty M μ 𝔾) ∧
      (∀ α ∈ adaptedPolicies M.A M.𝔽,
        ((∫ ω, M.r (α ω) ω ∂μ : ℝ) : EReal) = primalValue μ M.A M.𝔽 M.r →
          ((∫ ω, (M.r (α ω) ω - idealPenalty M μ 𝔾 (α ω) ω) ∂μ : ℝ) : EReal) =
            dualBound μ M.A 𝔾 M.r (idealPenalty M μ 𝔾)) ∧
      (𝔾 = perfectInfo →
        ∀ α ∈ adaptedPolicies M.A M.𝔽,
          ((∫ ω, M.r (α ω) ω ∂μ : ℝ) : EReal) = primalValue μ M.A M.𝔽 M.r →
            ∀ᵐ ω ∂μ, M.r (α ω) ω - idealPenalty M μ 𝔾 (α ω) ω = ∫ ω', M.r (α ω') ω' ∂μ) := by sorry

end InfoRelax.IdealPenalty
Source
Brown, Smith & Sun, Information Relaxations and Duality in Stochastic Dynamic Programs, Oper. Res. (Articles in Advance, 2010), DOI 10.1287/opre.1090.0796, p. 5, Theorem 2.3, eq. (11)
Human review
  • Endorsed by Shuze Chen · Oct 8, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 8, 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