Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 5 (Furstenberg–Kesten) — n⁻¹ log [Y₁⋯Y_n]_ij converges with probability one and in mean to a finite λ independent of i, j

Proved
KingmanSubadditive.PositiveMatrices.theorem_5

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

furstenberg-kestenlyapunov-exponentp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1random-matricessubadditive-ergodic-theorem

Let Y1,Y2,…Y_1,Y_2,\dotsY1​,Y2​,… be random k×kk\times kk×k matrices (k≥1k\ge1k≥1) on a probability space, and let

Xn=Y1Y2⋯Yn.X_n=Y_1Y_2\cdots Y_n .Xn​=Y1​Y2​⋯Yn​.

Suppose that

  1. the elements of the matrices YnY_nYn​ are strictly positive;
  2. their logarithms have finite expectations: E∣log⁡[Yn]ij∣<∞E|\log [Y_n]_{ij}|<\inftyE∣log[Yn​]ij​∣<∞ for all n≥1n\ge1n≥1 and all i,ji,ji,j;
  3. the sequence (Yn)(Y_n)(Yn​) is stationary (its joint law is invariant under the shift n↦n+1n\mapsto n+1n↦n+1).

Then there is a single finite random variable λ\lambdaλ with E∣λ∣<∞E|\lambda|<\inftyE∣λ∣<∞ such that for every (i,j)(i,j)(i,j) the limit

λ=lim⁡n→∞n−1log⁡[Xn]ij(2.2.2)\lambda=\lim_{n\to\infty} n^{-1}\log [X_n]_{ij}\qquad(2.2.2)λ=n→∞lim​n−1log[Xn​]ij​(2.2.2)

exists with probability one and in mean (E∣n−1log⁡[Xn]ij−λ∣→0E|n^{-1}\log [X_n]_{ij}-\lambda|\to0E∣n−1log[Xn​]ij​−λ∣→0); in particular it does not depend on iii or jjj.

This is (a variant of) the Furstenberg–Kesten theorem on products of random matrices; Kingman derives it as a corollary of his subadditive ergodic theorem. The quantity λ\lambdaλ is the top Lyapunov exponent of the product.

Formalization Note. One random variable serves every entry. The logarithm is applied only to strictly positive numbers because positivity of the YnY_nYn​ is a hypothesis; the paper's index 111 is 0 : Fin k.

Preamble
import Mathlib
import Definitions.Def_KingmanSubadditive_PositiveMatrices_Model
open MeasureTheory Filter Topology
Formal statement
namespace KingmanSubadditive.PositiveMatrices

/-- **Theorem 5** (Furstenberg–Kesten), p. 891 (Kingman, *Subadditive ergodic theory*, Ann.
Probab. 1(6):883–899 (1973)). Suppose that the elements of the random `k × k` matrices `Y_n` are
strictly positive, that their logarithms have finite expectations, and that the sequence
`(Y_n)` is stationary. Then the finite limit `λ = lim_{n→∞} n⁻¹ log [X_n]_ij`, (2.2.2), with
`X_n = Y₁ Y₂ ⋯ Y_n`, exists with probability one and in mean, and does not depend on `i` or `j`.

**Formalization Note.** One random variable `lam` (`λ` is a Lean keyword) serves every
`(i, j)`: that is "does not depend on i or j". "Finite" is `lam : Ω → ℝ`; "in mean" is
`E|n⁻¹ log [X_n]_ij − λ| → 0` with `λ` integrable. `Real.log` is the genuine logarithm because
every entry of `X_n`, `n ≥ 1`, is strictly positive. Stationarity is joint-law stationarity of
`(Y_n)_{n≥1}`, positivity holds for every outcome, `k ≥ 1`. -/
theorem theorem_5 {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω) [IsProbabilityMeasure P]
    {k : ℕ} [NeZero k] (Y : ℕ → Ω → Matrix (Fin k) (Fin k) ℝ) (hY : Hypotheses P Y) :
    ∃ lam : Ω → ℝ, Integrable lam P ∧ ∀ i j : Fin k,
      (∀ᵐ ω ∂P, Tendsto (fun n : ℕ => Real.log (X Y n ω i j) / n) atTop (𝓝 (lam ω))) ∧
      Tendsto (fun n : ℕ => ∫ ω, |Real.log (X Y n ω i j) / n - lam ω| ∂P) atTop (𝓝 0) := by sorry

end KingmanSubadditive.PositiveMatrices
Source
Kingman, Subadditive ergodic theory, Ann. Probab. 1(6):883–899 (1973), DOI 10.1214/aop/1176996798, p. 891, Theorem 5
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