Theorem 2.5.9 — Kronecker's lemma
ProvedDurrettProbability.kronecker_lemmaLet be a positive increasing sequence with , and let be real numbers. If the series converges, then
There is no probability in this statement — it is a lemma about real sequences — but it is what converts every theorem about convergence of random series into a theorem about averages, and therefore the bridge from section 2.5 to the strong law of large numbers. Applying the three-series theorem to and then Kronecker's lemma with turns " converges" into "", which is the law of large numbers.
The proof is summation by parts: writing and , the average becomes minus a weighted average of with weights summing to one, and both converge to the same limit.
Formalization Note "Increases to infinity" is split into positivity, monotonicity and divergence, since positivity is what makes the divisions meaningful and is used separately from the other two. Convergence of is convergence of the partial sums to a real limit, as in the rest of this mission, and not absolute convergence — the hypothesis would otherwise be much stronger than the book's and the lemma correspondingly weaker.
import Mathlib import Definitions.Def_DurrettProbability_Series open MeasureTheory ProbabilityTheory Filter
namespace DurrettProbability
theorem kronecker_lemma (a x : ℕ → ℝ) (hapos : ∀ n, 0 < a n) (hamono : Monotone a)
(hatop : Tendsto a atTop atTop)
(hconv : ∃ L : ℝ, Tendsto (fun N => ∑ n ∈ Finset.range N, x n / a n) atTop (nhds L)) :
Tendsto (fun n => (∑ m ∈ Finset.range n, x m) / a n) atTop (nhds 0) := by sorry
end DurrettProbabilityConfirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.