While this submission is a draft, it cannot be used by other submissions.

Birkhoff's pointwise ergodic theorem

Lax606786.BirkhoffErgodicTheorem · concepts/Lax606786/BirkhoffErgodicTheorem.lean · lax-606786

proven

Loading review…

Sign in with ORCID

Community review

Flags

Each flag is tied to a public ORCID identity and explains why this concept may be incorrect.

No flags have been submitted.

    Community review

    Flag this concept

    State precisely what appears incorrect. This explanation will be public under your ORCID name.

    No source line selected.

    Natural Language Statement

    Theorem

    Let σ\sigma be a measure-preserving transformation of a probability space (Ω,μ)(\Omega, \mu) and let f:Ω→Rf : \Omega \to \mathbb{R} be integrable. Let I\mathcal{I} be the σ\sigma-algebra of measurable sets AA with σ−1A=A\sigma^{-1} A = A. Then the Birkhoff averages converge almost everywhere and in L1(μ)L^1(\mu) to the conditional expectation of ff given I\mathcal{I}:

    1n∑k=0n−1f(σkω)→E[f∣I](ω)for μ-a.e. ω,∥1n∑k=0n−1f∘σk−E[f∣I]∥1→0.\frac1n \sum_{k=0}^{n-1} f(\sigma^k \omega) \to \mathbb{E}[f \mid \mathcal{I}](\omega) \quad \text{for } \mu\text{-a.e. } \omega, \qquad \Bigl\| \frac1n \sum_{k=0}^{n-1} f \circ \sigma^k - \mathbb{E}[f \mid \mathcal{I}] \Bigr\|_1 \to 0.

    When σ\sigma is ergodic, I\mathcal{I} is trivial and the limit is the constant ∫f dμ\int f \, d\mu.

    The almost-everywhere convergence is due to Birkhoff (1931). Convergence in mean goes back to von Neumann (1932), who proved it in L2L^2.

    Concept map
    1 concept
    100%
    Proven claimThis concept
    Evidence

    This concept declares 2 statements. Each proof establishes one of them relative to its assumptions.

    Lean source view on GitLab

    1import Mathlib.Dynamics.BirkhoffSum.Average
    2import Mathlib.Dynamics.Ergodic.Ergodic
    3import Mathlib.MeasureTheory.Function.ConditionalExpectation.Basic
    4import Mathlib.MeasureTheory.Integral.Bochner.Basic
    5import Mathlib.MeasureTheory.MeasurableSpace.Invariants
    6
    7/-!
    8---
    9title: Birkhoff's pointwise ergodic theorem
    10type: theorem
    11---
    12Let σ\sigma be a measure-preserving transformation of a probability space (Ω,μ)(\Omega, \mu) and let
    13f:Ω→Rf : \Omega \to \mathbb{R} be integrable. Let I\mathcal{I} be the σ\sigma-algebra of measurable
    14sets AA with σ−1A=A\sigma^{-1} A = A. Then the Birkhoff averages converge almost everywhere and in
    15L1(μ)L^1(\mu) to the conditional expectation of ff given I\mathcal{I}:
    161n∑k=0n−1f(σkω)→E[f∣I](ω)for μ-a.e. ω,∥1n∑k=0n−1f∘σk−E[f∣I]∥1→0.\frac1n \sum_{k=0}^{n-1} f(\sigma^k \omega) \to \mathbb{E}[f \mid \mathcal{I}](\omega) \quad \text{for } \mu\text{-a.e. } \omega, \qquad \Bigl\| \frac1n \sum_{k=0}^{n-1} f \circ \sigma^k - \mathbb{E}[f \mid \mathcal{I}] \Bigr\|_1 \to 0.
    17
    18
    19
    20When σ\sigma is ergodic, I\mathcal{I} is trivial and the limit is the constant ∫f dμ\int f \, d\mu.
    21
    22The almost-everywhere convergence is due to Birkhoff (1931). Convergence in mean goes back to von
    23Neumann (1932), who proved it in L2L^2.
    24-/
    25
    26namespace Lax606786.BirkhoffErgodicTheorem
    27
    28open MeasureTheory Filter Topology
    29
    30/-- The Birkhoff averages converge a.e. and in `L¹` to the conditional expectation of `f` given
    31the `σ`-algebra of `σ`-invariant sets. -/
    32axiom birkhoff {Ω : Type*} [MeasurableSpace Ω] (μ : Measure Ω) [IsProbabilityMeasure μ]
    33 (σ : Ω → Ω) (hσ : MeasurePreserving σ μ μ) (f : Ω → ℝ) (hf : Integrable f μ) :
    34 (∀ᵐ ω ∂μ, Tendsto (fun n : ℕ => birkhoffAverage ℝ σ f n ω) atTop
    35 (𝓝 ((μ[f | MeasurableSpace.invariants σ]) ω))) ∧
    36 Tendsto (fun n : ℕ =>
    37 ∫ ω, |birkhoffAverage ℝ σ f n ω - (μ[f | MeasurableSpace.invariants σ]) ω| ∂μ)
    38 atTop (𝓝 0)
    39
    40/-- For an ergodic transformation the Birkhoff averages of an integrable function converge almost
    41everywhere to its integral. -/
    42axiom birkhoff_of_ergodic {Ω : Type*} [MeasurableSpace Ω] (μ : Measure Ω)
    43 [IsProbabilityMeasure μ] (σ : Ω → Ω) (hσ : Ergodic σ μ) (f : Ω → ℝ) (hf : Integrable f μ) :
    44 ∀ᵐ ω ∂μ, Tendsto (fun n : ℕ => birkhoffAverage ℝ σ f n ω) atTop (𝓝 (∫ x, f x ∂μ))
    45
    46end Lax606786.BirkhoffErgodicTheorem
    47
    Show ProofShow Proof

    Discussion

    Ask a question or add context. Endorsements and structured flags are kept in the review panel above.

    Loading discussion…