The equilibrium distribution of an M/M/1 queue
ProvedKellyStochasticNetworks.mm1_equilibriumAn M/M/1 queue has a Poisson arrival stream of rate , a single server, and independent exponential service times of parameter . Writing for the number of customers present, including the one in service, is a Markov process on with transition rates
Suppose and write for the traffic intensity. Then the geometric distribution
satisfies the detailed balance equations for these rates, and its terms sum to . It is therefore the equilibrium distribution of the queue, and the queue is reversible.
The two conclusions are stated together because either alone is weaker than the claim: detailed balance is satisfied by every positive multiple of , while summing to says nothing about the dynamics.
Formalization Note The hypothesis is exactly what makes the geometric series converge, and it is the stability condition for the queue. Convergence and the value of the sum are asserted together.
import Mathlib import Definitions.Def_KellyStochasticNetworks_Balance import Definitions.Def_KellyStochasticNetworks_Migration
namespace KellyStochasticNetworks
theorem mm1_equilibrium (lam mu : ℝ) (hlam : 0 < lam) (hmu : lam < mu) :
DetailedBalance (fun j : ℕ => (1 - lam / mu) * (lam / mu) ^ j) (mm1Rates lam mu)
∧ HasSum (fun j : ℕ => (1 - lam / mu) * (lam / mu) ^ j) 1 := by sorry
end KellyStochasticNetworksConfirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.