Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

LEMMA — a sequence satisfying (5), (6), (7) tends to 0 at a rate depending only on a, b, c

Proved
VectorPayoffs.Convex.blackwell_lemma

by mikedeng1 · Sep 27, 2026 · Mathlib 0df444a (Lean v4.33.1)

almost-supermartingaleconvergence-ratep2o-batch-p200ap2o-gran-per-chapterp2o-plan-paperp2o-v1

Let a,b,c∈Ra,b,c\in\mathbb Ra,b,c∈R and ε>0\varepsilon>0ε>0. There is an integer N0N_0N0​, depending only on ε,a,b,c\varepsilon,a,b,cε,a,b,c, such that for every probability space and every sequence of measurable real random variables δ1,δ2,…\delta_1,\delta_2,\dotsδ1​,δ2​,… satisfying conditions (5), (6) and (7),

Prob{δn≥ε for some n≥N0}<ε.\mathrm{Prob}\{\delta_n\ge\varepsilon\text{ for some }n\ge N_0\}<\varepsilon .Prob{δn​≥ε for some n≥N0​}<ε.

Here (5) is E(δn∣δ1,…,δn−1)≤(1−2/n)δn−1+c/n2E(\delta_n\mid\delta_1,\dots,\delta_{n-1})\le(1-2/n)\delta_{n-1}+c/n^2E(δn​∣δ1​,…,δn−1​)≤(1−2/n)δn−1​+c/n2 whenever δn−1>0\delta_{n-1}>0δn−1​>0 (n≥2n\ge2n≥2), (6) is 0≤δn≤a0\le\delta_n\le a0≤δn​≤a (n≥1n\ge1n≥1), and (7) is ∣δn−δn−1∣≤b/n|\delta_n-\delta_{n-1}|\le b/n∣δn​−δn−1​∣≤b/n (n≥2n\ge2n≥2).

This is the probabilistic core of THEOREM 1: once the squared distance of the average payoff from SSS is shown to satisfy (5)–(7) with constants independent of the opponent's strategy, the uniform rate yields approachability.

Formalization Note The order of quantifiers is the paper's "an N0N_0N0​ depending only on ε,a,b,c\varepsilon,a,b,cε,a,b,c such that for any {δn}\{\delta_n\}{δn​}": N0N_0N0​ is chosen before the probability space (any space in Type) and the sequence. Conditions (6) and (7) are assumed only almost surely, which makes the statement stronger than a pointwise reading.

Preamble
import Mathlib
import Definitions.Def_VectorPayoffs_Convex_Recursion

open MeasureTheory
Formal statement
namespace VectorPayoffs.Convex

/-- Blackwell (1956), §2, p. 4, LEMMA: a sequence satisfying (5), (6), (7) tends to `0` with
probability `1` at a rate depending only on `a, b, c`: `N₀` is chosen before the probability
space and the sequence. -/
theorem blackwell_lemma (a b c ε : ℝ) (hε : 0 < ε) :
    ∃ N₀ : ℕ, ∀ (Ω : Type) [MeasurableSpace Ω] (μ : Measure Ω) [IsProbabilityMeasure μ]
      (δ : ℕ → Ω → ℝ), (∀ n, 1 ≤ n → Measurable (δ n)) → SatisfiesRecursion a b c μ δ →
        (μ {ω | ∃ n, N₀ ≤ n ∧ 1 ≤ n ∧ ε ≤ δ n ω}).toReal < ε := by sorry

end VectorPayoffs.Convex
Source
Blackwell, An analog of the minimax theorem for vector payoffs, Pacific J. Math. 6(1), 1956, p. 4, LEMMA (with displays (5), (6), (7) on the same page)
Human review
  • Endorsed by Shuze Chen · Sep 27, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Sep 27, 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