Eq. (4.26) — the marginal-probability recursion of a closed Jackson network
ProvedQueueingFundamentals.Networks.marginal_recursionConsider a closed Jackson network of nodes, each with a single exponential server of rate , and an irreducible routing matrix with rows summing to one. For each population let be the steady-state distribution of the -customer network (the probability solution of the balance equations (4.14)), let
be the marginal probability of customers at node , and let
be the throughput of node . Then and
The recursion is the closed-network analogue of for the queue: combined with mean-value analysis, it yields every nodal marginal distribution population by population, without computing the normalizing constant.
Formalization Note is not a free parameter: it is the throughput computed from , as in the book's proof (p.209). No traffic-equation solution appears in the statement; its existence follows from the irreducibility of .
import Mathlib import Definitions.Def_QueueingFundamentals_Networks_ClosedJackson
namespace QueueingFundamentals.Networks
/-- Eq. (4.26), p.204 (proved pp.207–209): for a closed Jackson network of `k ≥ 1` single-server
nodes (rates `μ_i > 0`, irreducible routing `R`) and its steady-state distributions `p_N`
(`N = 0, 1, 2, …`), the marginal probabilities `p_i(n, N)` satisfy `p_i(0, 0) = 1` and
`p_i(n, N) = (λ_i(N)/μ_i) p_i(n − 1, N − 1)` for `n, N ≥ 1`, where
`λ_i(N) = Pr{server busy at node i} · μ_i` is the throughput of node `i`. -/
theorem marginal_recursion {k : ℕ} [NeZero k] (mu : Fin k → ℝ) (R : Fin k → Fin k → ℝ)
(hmu : ∀ i, 0 < mu i) (hR : IsRoutingMatrix R) (hirr : IsIrreducible R)
(p : ℕ → (Fin k → ℕ) → ℝ) (hp : ∀ N, IsClosedSteadyState mu R N (p N)) (i : Fin k) :
marginal 0 (p 0) i 0 = 1 ∧
∀ n N : ℕ, 1 ≤ n → 1 ≤ N →
marginal N (p N) i n =
throughput mu N (p N) i / mu i * marginal (N - 1) (p (N - 1)) i (n - 1) := by sorry
end QueueingFundamentals.Networks
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.