Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem B (Anscombe 1952) — the central limit theorem for sums with a random number m_t of terms, m_t/ψ(t) → 1 in probability

Proved
SmithRegenerative.CLT.theorem_B

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

central-limit-theoremlimit-theoremp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1random-sums

Let ψ\psiψ be an unbounded, non-decreasing real function of a real variable ttt. For each ttt let mtm_tmt​ be a random variable with values in the positive integers, such that mt/ψ(t)→1m_t/\psi(t) \to 1mt​/ψ(t)→1 in probability as t→∞t \to \inftyt→∞. Let y1,y2,…y_1, y_2, \dotsy1​,y2​,… be independent, identically distributed real random variables on the same probability space with

Eyi=0,var⁡yi=σ2,σ>0.E y_i = 0, \qquad \operatorname{var} y_i = \sigma^2, \qquad \sigma > 0.Eyi​=0,varyi​=σ2,σ>0.

Then for every real α\alphaα, as t→∞t \to \inftyt→∞,

P{∑i=1mtyiσ [ψ(t)]1/2≤α}→Φ(α)=∫−∞αe−θ2/22π dθ.P\left\{ \frac{\sum_{i=1}^{m_t} y_i}{\sigma\,[\psi(t)]^{1/2}} \le \alpha \right\} \to \Phi(\alpha) = \int_{-\infty}^{\alpha} \frac{e^{-\theta^2/2}}{\sqrt{2\pi}}\,\mathrm{d}\theta.P{σ[ψ(t)]1/2∑i=1mt​​yi​​≤α}→Φ(α)=∫−∞α​2π​e−θ2/2​dθ.

No independence between the random index mtm_tmt​ and the summands is assumed; this is what makes the theorem applicable to mt=nt+1m_t = n_t + 1mt​=nt​+1, the number of renewal cycles begun by time ttt, which is determined by the cycles themselves. It is the main tool of Smith's central limit theorems.

Formalization Note σ>0\sigma > 0σ>0 is implicit in the paper, which divides by σ\sigmaσ. Φ\PhiΦ is the distribution function of the standard normal law. The sequence starts at y1y_1y1​.

Preamble
import Mathlib
import Definitions.Def_SmithRegenerative_CLT_CumulativeProcess

open MeasureTheory ProbabilityTheory Filter Topology
Formal statement
namespace SmithRegenerative.CLT

/-- Smith (1955), Theorem B, pp. 28–29 (Anscombe 1952, in Smith's form): "Let `ψ(t)` be an
unbounded, non-decreasing function of a real variable `t`. Let `m_t` be a proper random variable
taking positive integer values, and such that `m_t/ψ(t) → 1`, in probability. Let `{y_i}` be a
sequence of identically distributed, independent random variables, such that `Ey_i = 0`,
`var y_i = σ²` (5·4·1). Then, as `t → ∞`, `P{Σ₁^{m_t} y_i/(σ[ψ(t)]^{1/2}) ≤ α} → Φ(α)` (5·4·2)."

**Formalization Note** `σ > 0` is implicit in the paper (it divides by `σ`). No independence
between `m_t` and the `y_i` is assumed. `y 0` is unused; the sequence is `y 1, y 2, …`.
`Φ` is the distribution function of the standard normal law, `cdf (gaussianReal 0 1)`, and the
convergence is asserted for every real `α`. -/
theorem theorem_B {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω) [IsProbabilityMeasure P]
    (ψ : ℝ → ℝ) (hψ_mono : Monotone ψ) (hψ_unbdd : ¬ BddAbove (Set.range ψ))
    (m : ℝ → Ω → ℕ) (hm_meas : ∀ t, Measurable (m t)) (hm_pos : ∀ t ω, 1 ≤ m t ω)
    (hm_ratio : TendstoInMeasure P (fun t ω => (m t ω : ℝ) / ψ t) atTop (fun _ => 1))
    (y : ℕ → Ω → ℝ) (hy_indep : iIndepFun (fun i => y (i + 1)) P)
    (hy_ident : ∀ i, IdentDistrib (y (i + 1)) (y 1) P P)
    (hy_L2 : MemLp (y 1) 2 P) (hy_mean : ∫ ω, y 1 ω ∂P = 0)
    (σ : ℝ) (hσ_pos : 0 < σ) (hy_var : variance (y 1) P = σ ^ 2) :
    ∀ α : ℝ, Tendsto
      (fun t : ℝ => P.real {ω | (∑ i ∈ Finset.Icc 1 (m t ω), y i ω) / (σ * Real.sqrt (ψ t)) ≤ α})
      atTop (𝓝 (cdf (gaussianReal 0 1) α)) := by sorry

end SmithRegenerative.CLT
Source
Smith, Regenerative stochastic processes, Proc. R. Soc. Lond. A 232(1188):6–31 (1955), DOI 10.1098/rspa.1955.0198, pp. 28–29, Theorem B (5·4·1)–(5·4·2), after Anscombe (1952)
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