Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 6.1, p. 29 — agnostic Q-NPG bound from transfer error, excess risk and conditioning

Proved
PolicyGradTheory.QNPG.qnpg_agnostic_bound

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

agnostic-learninglog-linear-policyp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1q-npgtransfer-error

Fix finite state and action sets, a discounted MDP with rewards in [0,1], a start distribution ρ, a state-action fitting distribution ν, and any comparator policy π⋆. Let πθ be the log-linear class with feature norms at most B. Q-NPG starts from θ(0)=0\theta^{(0)}=0θ(0)=0 and updates with random directions w(t)w^{(t)}w(t) in the radius-W ball and step size η=2log⁡∣A∣/(B2W2T)\eta=\sqrt{2\log|A|/(B^2W^2T)}η=2log∣A∣/(B2W2T)​. At each t<Tt<Tt<T, let w⋆(t)w_\star^{(t)}w⋆(t)​ be an exact radius-W minimizer of the Q-prediction loss under the on-policy measure d(t)d^{(t)}d(t).

Suppose the relative covariance bound with constant κ holds between d⋆=dρπ⋆×UnifAd^\star=d^{\pi^\star}_\rho\times\mathrm{Unif}_Ad⋆=dρπ⋆​×UnifA​ and ν. Suppose the expected excess risk is at most εstat and the expected transfer loss L(w⋆(t);θ(t),d⋆)L(w_\star^{(t)};\theta^{(t)},d^\star)L(w⋆(t)​;θ(t),d⋆) is at most εbias at every t<Tt<Tt<T. Then

E ⁣[min⁡t<T{Vπ⋆(ρ)−Vπθ(t)(ρ)}]≤BW1−γ2log⁡∣A∣T+4∣A∣κεstat(1−γ)3+4∣A∣εbias1−γ.\mathbb E\!\left[\min_{t<T}\{V^{\pi^\star}(\rho)-V^{\pi_{\theta^{(t)}}}(\rho)\}\right] \le \frac{BW}{1-\gamma}\sqrt{\frac{2\log|A|}{T}} +\sqrt{\frac{4|A|\kappa\varepsilon_{\rm stat}}{(1-\gamma)^3}} +\frac{\sqrt{4|A|\varepsilon_{\rm bias}}}{1-\gamma}.E[t<Tmin​{Vπ⋆(ρ)−Vπθ(t)​(ρ)}]≤1−γBW​T2log∣A∣​​+(1−γ)34∣A∣κεstat​​​+1−γ4∣A∣εbias​​​.

The result guarantees competition with an arbitrary comparison policy; optimality of π⋆ is not assumed.

Formalization Note The finite setting and reward range come from §3. Nonempty A and positive B, W and T make the uniform policy, step size and minimum well-defined. Random directions and exact minimizers are measurable, which makes these bounded-loss expectations integrable. κ is a nonnegative quadratic-form upper bound, which includes the finite coefficient of Assumption 6.2. The minimum is taken separately for each random run before expectation.

Preamble
import Mathlib
import Definitions.Def_PolicyGradTheory_QNPG_Setting
Formal statement
namespace PolicyGradTheory.QNPG

open FoundationsML.ReinforcementLearning MeasureTheory

/-- Theorem 6.1, p. 29: agnostic Q-NPG for a log-linear policy class. -/
theorem qnpg_agnostic_bound {S A : Type} [Fintype S] [DecidableEq S]
    [Fintype A] [DecidableEq A] [Nonempty A] (d : ℕ)
    (P : S → A → S → ℝ) (r : S → A → ℝ) (γ : ℝ)
    (hM : PolicyGradTheory.ProjGA.IsFiniteMDP P r γ)
    (ρ : S → ℝ) (hρ : PolicyGradTheory.ProjGA.IsDist ρ)
    (ν : S × A → ℝ) (hν : PolicyGradTheory.ProjGA.IsDist ν)
    (πstar : S → A → ℝ) (hπstar : IsPolicy πstar)
    (φ : S → A → EuclideanSpace ℝ (Fin d))
    (B : ℝ) (hB : 0 < B) (hφ : ∀ s a, ‖φ s a‖ ≤ B)
    (W : ℝ) (hW : 0 < W)
    (κ : ℝ) (hκnonneg : 0 ≤ κ)
    (hκ : ∀ v, quadForm φ (dstar P γ ρ πstar) v ≤ κ * quadForm φ ν v)
    (T : ℕ) (hT : 0 < T)
    (η : ℝ) (hη : η = Real.sqrt (2 * Real.log (Fintype.card A : ℝ) /
      (B ^ 2 * W ^ 2 * T)))
    {Ω : Type*} [MeasurableSpace Ω] (ℙ : Measure Ω) [IsProbabilityMeasure ℙ]
    (w wstar : ℕ → Ω → EuclideanSpace ℝ (Fin d))
    (hwm : ∀ t < T, Measurable (w t))
    (hwstarm : ∀ t < T, Measurable (wstar t))
    (hwW : ∀ t < T, ∀ ω, ‖w t ω‖ ≤ W)
    (hwstar : ∀ t < T, ∀ ω,
      ‖wstar t ω‖ ≤ W ∧
        ∀ v, ‖v‖ ≤ W →
          qLoss P r γ φ (wstar t ω) (qnpgParams η w t ω)
            (onPolicyMeasure (logLinearPolicy φ (qnpgParams η w t ω)) P γ ν) ≤
          qLoss P r γ φ v (qnpgParams η w t ω)
            (onPolicyMeasure (logLinearPolicy φ (qnpgParams η w t ω)) P γ ν))
    (εstat εbias : ℝ)
    (hstat : ∀ t < T,
      (∫ ω, (qLoss P r γ φ (w t ω) (qnpgParams η w t ω)
                (onPolicyMeasure (logLinearPolicy φ (qnpgParams η w t ω)) P γ ν) -
              qLoss P r γ φ (wstar t ω) (qnpgParams η w t ω)
                (onPolicyMeasure (logLinearPolicy φ (qnpgParams η w t ω)) P γ ν)) ∂ℙ) ≤ εstat)
    (hbias : ∀ t < T,
      (∫ ω, qLoss P r γ φ (wstar t ω) (qnpgParams η w t ω)
        (dstar P γ ρ πstar) ∂ℙ) ≤ εbias) :
    (∫ ω, (Finset.range T).inf' (Finset.nonempty_range_iff.mpr (Nat.ne_of_gt hT))
      (fun t => PolicyGradTheory.ProjGA.valueAt πstar P r γ ρ -
        PolicyGradTheory.ProjGA.valueAt (logLinearPolicy φ (qnpgParams η w t ω)) P r γ ρ) ∂ℙ) ≤
      B * W / (1 - γ) * Real.sqrt (2 * Real.log (Fintype.card A : ℝ) / T) +
      Real.sqrt (4 * (Fintype.card A : ℝ) * κ * εstat / (1 - γ) ^ 3) +
      Real.sqrt (4 * (Fintype.card A : ℝ) * εbias) / (1 - γ) := by sorry

end PolicyGradTheory.QNPG
Source
arXiv:1908.00261v5, Theorem 6.1, p. 29; Assumptions 6.1–6.2, pp. 28–29
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