The uncapacitated loss network: independent Poisson equilibrium
ProvedKellyStochasticNetworks.loss_network_uncapacitatedConsider a loss network with fixed routing in which every link has infinite capacity, . Arriving calls are then never blocked: a call on route simply arrives at rate , stays for an exponentially distributed time of unit mean, and leaves. The state is a linear migration process with transition rates
that is, the open migration process of Chapter 2 with , and .
Its equilibrium distribution is a product of independent Poisson distributions,
and the process is in fact reversible. Concretely, for this satisfies the equilibrium equations for those rates, sums to over , and satisfies the detailed balance equations.
Because the capacities are infinite the individual routes are independent, which is what makes the product form here elementary. Reversibility is the extra ingredient: it is what lets Lemma 3.4 be applied to the truncation of this process to the feasible set of a network with finite capacities, producing the exact equilibrium distribution (3.3).
Formalization Note The rates are written as the open migration rates published in mission II of this series, instantiated at , , , so the claim is literally that this loss network is that migration process. The traffic equations (2.2) are then solved by , and , so the first two conclusions are an instance of the product-form theorem of mission II; the third, detailed balance, is not and has to be checked directly.
import Mathlib import Definitions.Def_KellyStochasticNetworks_Balance import Definitions.Def_KellyStochasticNetworks_Erlang import Definitions.Def_KellyStochasticNetworks_Migration import Definitions.Def_KellyStochasticNetworks_LossNetwork
namespace KellyStochasticNetworks
theorem loss_network_uncapacitated {R : ℕ} (ν : Fin R → ℝ) (hν : ∀ r, 0 < ν r) :
FullBalance
(fun n : Fin R → ℕ => ∏ r, Real.exp (-ν r) * (ν r ^ n r / (Nat.factorial (n r) : ℝ)))
(openMigrationRates (fun _ _ => (0 : ℝ)) (fun _ => 1) ν (fun _ m => (m : ℝ)))
∧ HasSum
(fun n : Fin R → ℕ => ∏ r, Real.exp (-ν r) * (ν r ^ n r / (Nat.factorial (n r) : ℝ))) 1
∧ DetailedBalance
(fun n : Fin R → ℕ => ∏ r, Real.exp (-ν r) * (ν r ^ n r / (Nat.factorial (n r) : ℝ)))
(openMigrationRates (fun _ _ => (0 : ℝ)) (fun _ => 1) ν (fun _ m => (m : ℝ))) := by sorry
end KellyStochasticNetworksConfirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.