Theorem 12 — given N(t) = n, the demand epochs are order statistics of n i.i.d. draws from F = m(x)/m(t)
ProvedServiceParts.NonstatPalm.arrival_times_order_statisticsConsider the single-location model with nonstationary Poisson demand of rate , mean function , and time-dependent resupply times. Fix with and let be the distribution function
Let be the probability measure on with distribution function , and let be independent with law , with order statistics . Then, for every and every measurable ,
That is, conditionally on , the demand epochs in have the same distribution as the order statistics of independent random variables with distribution function .
This is the nonstationary counterpart of the uniform order-statistics property of a Poisson process, and it is the step on which the proof of Theorem 13 rests.
Formalization Note The book writes only for ; for is added, since demands occur at nonnegative times. The hypothesis makes well defined; when there are almost surely no demands in and the book's is . The measure is quantified universally over probability measures with distribution function ; there is exactly one. The order statistics are obtained by sorting the vector .
import Mathlib import Definitions.Def_ServiceParts_NonstatPalm_ResupplySystem open MeasureTheory ProbabilityTheory
namespace ServiceParts.NonstatPalm
theorem arrival_times_order_statistics {Ω : Type*} [MeasurableSpace Ω] {P : Measure Ω}
[IsProbabilityMeasure P] {E : Type*} [MeasurableSpace E] (S : ResupplySystem Ω P E)
{t : ℝ} (hm : 0 < S.meanFn t) (n : ℕ)
(μ : Measure ℝ) [IsProbabilityMeasure μ]
(hμ : ∀ x : ℝ, μ (Set.Iic x) = ENNReal.ofReal (S.arrivalCdf t x))
{A : Set (Fin n → ℝ)} (hA : MeasurableSet A) :
P ({ω | S.demandCount t ω = n} ∩ {ω | (fun i : Fin n => S.epoch (i : ℕ) ω) ∈ A}) =
P {ω | S.demandCount t ω = n} *
Measure.pi (fun _ : Fin n => μ) {x : Fin n → ℝ | x ∘ Tuple.sort x ∈ A} := by sorry
end ServiceParts.NonstatPalm
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.