Eq. (6.18) — the M/D/c generating function
ProvedQueueingFundamentals.GG1.mdc_pgfConsider the M/D/c queue with servers, the constant service time as the unit of time, and Poisson arrivals at rate per unit. If , a steady-state distribution solving (6.17) exists. For every steady-state distribution, with and , and every complex with ,
This expresses the queue-size generating function through the unknown probabilities , which the roots of the denominator then determine.
Formalization Note (6.18) is stated with its denominator cleared, , on the whole closed disk; this is equivalent to the quotient wherever the denominator is nonzero and additionally says that the numerator vanishes at the zeros of the denominator. The existence half under is the book's steady-state assumption ( after rescaling time).
import Mathlib import Definitions.Def_QueueingFundamentals_GG1_MDc
namespace QueueingFundamentals.GG1
/-- Eq. (6.18) (p.295) for the M/D/c queue (`c` servers, constant service time taken as the unit of
time, Poisson arrivals at rate `λ > 0` per unit). Under the steady-state condition `λ < c` a
probability solution of (6.17) exists; and for every probability solution `p` of (6.17) and every
`|z| ≤ 1`,
`P(z)(1 − z^c e^{λ(1−z)}) = ∑_{n=0}^{c} p_n z^n − P_c z^c = ∑_{n=0}^{c−1} p_n (z^n − z^c)`,
the cleared-denominator form of `P(z) = (∑_{n=0}^{c} p_n z^n − P_c z^c)/(1 − z^c e^{λ(1−z)})`. -/
theorem mdc_pgf (lam : ℝ) (hlam : 0 < lam) (c : ℕ) (hc : 1 ≤ c) :
((lam < c) → ∃ p : ℕ → ℝ, IsMDcStationary lam c p) ∧
∀ p : ℕ → ℝ, IsMDcStationary lam c p → ∀ z : ℂ, ‖z‖ ≤ 1 →
QueueingFundamentals.MG1.pgf p z * (1 - z ^ c * Complex.exp ((lam : ℂ) * (1 - z))) =
(∑ n ∈ Finset.range (c + 1), (p n : ℂ) * z ^ n) - (cumProb p c : ℂ) * z ^ c ∧
QueueingFundamentals.MG1.pgf p z * (1 - z ^ c * Complex.exp ((lam : ℂ) * (1 - z))) =
∑ n ∈ Finset.range c, (p n : ℂ) * (z ^ n - z ^ c) := by sorry
end QueueingFundamentals.GG1
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.