Maximal bounds for centered independent partial sums #
Independent integrable centered increments yield a martingale of inclusive partial sums.
Its second moment is the sum of increment second moments, giving Doob's explicit maximal bound.
This complements the tail-based walk inequalities in RealWalkMaximal without duplicating them.
theorem
BGW.Probability.martingale_centered_partialSums
{Ω : Type u_1}
{mΩ : MeasurableSpace Ω}
{P : MeasureTheory.Measure Ω}
[MeasureTheory.IsProbabilityMeasure P]
{X : ℕ → Ω → ℝ}
(hm : ∀ (i : ℕ), MeasureTheory.StronglyMeasurable (X i))
(hi : ProbabilityTheory.iIndepFun X P)
(hint : ∀ (i : ℕ), MeasureTheory.Integrable (X i) P)
(hzero : ∀ (i : ℕ), ∫ (ω : Ω), X i ω ∂P = 0)
:
MeasureTheory.Martingale (fun (n : ℕ) (ω : Ω) => ∑ i ∈ Finset.range (n + 1), X i ω)
(MeasureTheory.Filtration.natural X hm) P
Independent integrable centered increments give a martingale of inclusive partial sums.
theorem
BGW.Probability.integral_partialSum_sq_eq_sum
{Ω : Type u_1}
{mΩ : MeasurableSpace Ω}
{P : MeasureTheory.Measure Ω}
[MeasureTheory.IsProbabilityMeasure P]
{X : ℕ → Ω → ℝ}
(hmem : ∀ (i : ℕ), MeasureTheory.MemLp (X i) 2 P)
(hi : ProbabilityTheory.iIndepFun X P)
(hzero : ∀ (i : ℕ), ∫ (ω : Ω), X i ω ∂P = 0)
(s : Finset ℕ)
:
The centered second moment of independent partial sums is the sum of their variances.
theorem
BGW.Probability.measure_partialSum_maximal_le
{Ω : Type u_1}
{mΩ : MeasurableSpace Ω}
{P : MeasureTheory.Measure Ω}
[MeasureTheory.IsProbabilityMeasure P]
{X : ℕ → Ω → ℝ}
(hm : ∀ (i : ℕ), MeasureTheory.StronglyMeasurable (X i))
(hi : ProbabilityTheory.iIndepFun X P)
(hmem : ∀ (i : ℕ), MeasureTheory.MemLp (X i) 2 P)
(hzero : ∀ (i : ℕ), ∫ (ω : Ω), X i ω ∂P = 0)
(m : ℕ)
{r : ℝ}
(hr : 0 < r)
:
A second-moment maximal bound for independent centered increments, allowing one extra terminal increment in the upper bound.