Eq. (4.22) — the marginal distribution at the last node via Buzen's functions
ProvedQueueingFundamentals.Networks.last_node_marginalbuzen-algorithmclosed-networkmarginal-distributionp2o-batch-b23bp2o-gran-per-chapterp2o-plan-bookp2o-v1queueing
Consider a closed network of nodes with servers at node and , whose -customer joint distribution is the product form (4.17),
with and as in (4.19)–(4.20). Then the marginal distribution of the number of customers at the last node is
Together with Buzen's recursion this yields a node's marginal distribution without summing over the joint distribution; any node can be made the last one by relabelling.
Formalization Note The network has nodes in Lean (Fin (k + 1)), the last one being Fin.last k, so the book's is gBuzen … k.
Preamble
import Mathlib import Definitions.Def_QueueingFundamentals_Networks_ClosedJackson
Formal statement
namespace QueueingFundamentals.Networks
/-- Eq. (4.22), p.199: in a closed network of `k + 1` nodes (the last node is node `k + 1` of the book,
index `Fin.last k`) with `c_i ≥ 1` servers at node `i` and `ρ_i > 0`, under the product form (4.17)
the marginal distribution of the last node is `p_{k+1}(n) = f_{k+1}(n) g_k(N − n) / G(N)` for
`n = 0, 1, …, N`. -/
theorem last_node_marginal {k : ℕ} (rho : Fin (k + 1) → ℝ) (c : Fin (k + 1) → ℕ)
(hrho : ∀ i, 0 < rho i) (hc : ∀ i, 1 ≤ c i) (N n : ℕ) (hn : n ≤ N) :
marginal N (productForm (buzenFactor rho c) N) (Fin.last k) n =
buzenFactor rho c (Fin.last k) n * gBuzen (buzenFactor rho c) k (Nat.le_succ k) (N - n) /
normConst (buzenFactor rho c) N := by sorry
end QueueingFundamentals.Networks
Source
Gross, Shortle, Thompson & Harris, Fundamentals of Queueing Theory, 4th ed., Wiley 2008, DOI 10.1002/9781118625651, p.199, Eq. (4.22)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.