Theorem 6, p. 893 — for stationary (Y_n) in a Banach algebra with E(log‖Y₁‖)⁺ < ∞, n⁻¹ log‖Y₁⋯Y_n‖ → ξ a.s. and E(ξ) = lim n⁻¹E log‖Y₁⋯Y_n‖
ProvedKingmanSubadditive.BanachAlgebra.theorem_6Let be a real or complex Banach algebra and let be a stationary sequence of random elements of on a probability space . Suppose that
Then the limit
exists in with probability one, and
The quantity is a stochastic analogue of the spectral radius: if every equals a fixed element , then is the spectral radius of by Gelfand's formula. For matrices with any matrix norm the theorem is the Furstenberg–Kesten theorem on products of random matrices.
Formalization Note is extended-real valued () and all expectations are extended expectations . The conclusion asserts: is a measurable -valued random variable; almost surely and in ; ; and in , which asserts both that the limit exists and that it equals . Stationarity is equality of joint laws on with the product -algebra; each is strongly measurable and need not be separable. is a complete unital normed ring and a normed algebra over or .
import Mathlib import Definitions.Def_KingmanSubadditive_BanachAlgebra_Process import Definitions.Def_KingmanSubadditive_BanachAlgebra_RandomProduct
namespace KingmanSubadditive.BanachAlgebra
open MeasureTheory Filter Topology
/-- **Theorem 6, §2.3, p. 893** (Kingman, *Subadditive ergodic theory*, Ann. Probab.
1(6):883–899 (1973), DOI 10.1214/aop/1176996798). "Let `(Y_n)` be a stationary sequence taking
values in a (real or complex) Banach algebra `𝔅`, and suppose that (2.3.3)
`E{(log ‖Y₁‖)⁺} < ∞`. Then (2.3.4) `ξ = lim_{n→∞} n⁻¹ log ‖Y₁ Y₂ ⋯ Y_n‖` exists in
`−∞ ≤ ξ < ∞` with probability one, and (2.3.5) `E(ξ) = lim_{n→∞} n⁻¹ E{log ‖Y₁ Y₂ ⋯ Y_n‖}`."
**Formalization Note.** `log ‖·‖` is the extended-real logarithm (`log ‖0‖ = −∞`), so
`log ‖Y₁ ⋯ Y_n‖`, `ξ` and all means are in `EReal`; expectations are the extended expectation
`eMean` (`∫ f⁺ − ∫ f⁻`, possibly `−∞`). The conclusion asserts: `ξ` is a measurable
`EReal`-valued random variable; almost surely `ξ < +∞` and `n⁻¹ log ‖Y₁ ⋯ Y_n‖ → ξ` in `[−∞, ∞]`;
`E(ξ⁺) < ∞` (so `E(ξ)` is defined); and the sequence `n⁻¹ E{log ‖Y₁ ⋯ Y_n‖}` converges to `E(ξ)`
in `[−∞, ∞]` (both existence of the limit and its value). The term at `n = 0` does not affect the
limits. The product is ordered `Y₁ Y₂ ⋯ Y_n`. Each `Y_n` (`n ≥ 1`) is strongly measurable and the
sequence is stationary in the joint-law sense (`IsStationarySeq`); `𝔅` need not be separable. -/
theorem theorem_6 {𝕜 𝔅 Ω : Type*} [RCLike 𝕜] [NormedRing 𝔅]
[NormedAlgebra 𝕜 𝔅] [CompleteSpace 𝔅] [MeasurableSpace 𝔅] [BorelSpace 𝔅]
[MeasurableSpace Ω] (P : Measure Ω) [IsProbabilityMeasure P] (Y : ℕ → Ω → 𝔅)
(hY : ∀ n : ℕ, 1 ≤ n → StronglyMeasurable (Y n)) (hstat : IsStationarySeq P Y)
(hlog : ∫⁻ ω, (logNorm (Y 1 ω)).toENNReal ∂P < ⊤) :
∃ ξ : Ω → EReal, Measurable ξ ∧
(∀ᵐ ω ∂P, ξ ω ≠ ⊤ ∧
Tendsto (fun n : ℕ => (((n : ℝ)⁻¹ : ℝ) : EReal) * logNorm (prodRange Y 0 n ω))
atTop (𝓝 (ξ ω))) ∧
∫⁻ ω, (ξ ω).toENNReal ∂P < ⊤ ∧
Tendsto (fun n : ℕ => (((n : ℝ)⁻¹ : ℝ) : EReal) *
eMean P (fun ω => logNorm (prodRange Y 0 n ω)))
atTop (𝓝 (eMean P ξ)) := by sorry
end KingmanSubadditive.BanachAlgebra
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.