§1.2, p. 246 — the elementary renewal theorem: H(t)/t → μ₁⁻¹ as t → ∞ (μ₁ ≤ ∞, limit 0 if μ₁ = ∞)
ProvedSmithRenewal.Elementary.elementary_renewal_theoremLet be a renewal process: independent, identically distributed, non-negative random variables that do not vanish with probability one. Let , let be the number of renewals with , let be the renewal function and the mean lifetime. Then
where the limit is interpreted as zero if .
The elementary renewal theorem is the first-order asymptotics of the renewal function: in the long run renewals occur at rate on average. It underlies long-run average cost and reward computations for renewal and regenerative models in inventory, maintenance and queueing.
Formalization Note and are lower integrals in , the ratio is formed in , and the limit is the inverse in , where is exactly the paper's convention. No hypothesis on , no second moment and no non-lattice condition is assumed.
import Mathlib import Definitions.Def_SmithRenewal_Elementary_RenewalProcess open MeasureTheory ProbabilityTheory Filter Topology open scoped ENNReal
namespace SmithRenewal.Elementary
/-- The elementary renewal theorem (Smith, *Renewal Theory and Its Ramifications*, J. R. Statist. Soc. B 20(2):243–283 (1958), §1.2, p. 246, unnumbered display): "H(t)/t → μ₁⁻¹ as
t → ∞, where μ₁ = EX_i ≤ ∞, and the limit μ₁⁻¹ is interpreted as zero if μ₁ = ∞."
Formalization Note: `H(t) = E N_t` and `μ₁ = E X_i` are lower integrals in `[0, ∞]`, the ratio
is `H(t) / ofReal t` in `[0, ∞]` and the limit `μ₁⁻¹` is the inverse in `[0, ∞]`, so
`∞⁻¹ = 0` is exactly the paper's convention. No hypothesis on `μ₁` is made (`μ₁ > 0` follows
from `P {X_i = 0} < 1`). -/
theorem elementary_renewal_theorem {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω)
[IsProbabilityMeasure P] (X : ℕ → Ω → ℝ) (hX : IsRenewalProcess P X) :
Tendsto (fun t : ℝ => H X P t / ENNReal.ofReal t) atTop (𝓝 (mu1 X P)⁻¹) := by sorry
end SmithRenewal.Elementary
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.