Proposition 2.3, p. 712 — with all service times 1 in an IS model, N(t) = A(t) − A(t − 1)
ProvedWhittEfficiency.IS.proposition_2_3Consider an infinite-server model in which every customer has service time exactly . Let be the number of arrivals in and the number of busy servers at time , where an arrival at time is counted and a departure at time is not. Then, for every realization of the arrival epochs with only finitely many arrivals in ,
The statement is pathwise and holds for any arrival process: a customer arriving at or before has left by time , and every customer arriving in is still present. It reduces the busy-server count of the M/D/∞ model to an increment of the arrival counting process.
Formalization Note The arrival epochs are with all , so (countingProcess, which counts with ) is the number of arrivals in . The hypothesis that only finitely many epochs lie in is added because the counts are Set.ncard, which returns on an infinite set; it holds almost surely for Poisson arrivals. The identity is stated in to avoid truncated natural-number subtraction.
import Mathlib import Definitions.Def_WhittEfficiency_IS_Model open MeasureTheory ProbabilityTheory QueueingFundamentals.Foundations
namespace WhittEfficiency.IS
/-- Whitt 1992, Proposition 2.3, p. 712: if all service times are `1` in an IS model, then
`N(t) = A(t) − A(t − 1)`, where `A(t)` counts the arrivals in `(0, t]`. Pathwise; the arrival
epochs are positive (`hpos`) and only finitely many arrive by `t` (`hfin`, which makes the counts
genuine cardinalities rather than the junk value of `Set.ncard` on an infinite set). -/
theorem proposition_2_3 {Ω : Type*} (T : ℕ → Ω → ℝ) (ω : Ω) (t : ℝ)
(hpos : ∀ i, 0 < T i ω)
(hfin : {n : ℕ | 1 ≤ n ∧ arrivalTime T n ω ≤ t}.Finite) :
(busyCount T (fun _ _ => 1) t ω : ℤ) = countingProcess T t ω - countingProcess T (t - 1) ω := by sorry
end WhittEfficiency.IS
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.