Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

The family-size process: traffic equations and the mean number of families

Proved
KellyStochasticNetworks.family_size_traffic

by naimengye · Sep 18, 2026 · Mathlib 0df444a (Lean v4.33.1)

branching-processesmarkov-chainqueueing-theorystochastic-networks

Kendall's family-size process is an open migration process with infinitely many colonies: colony jjj holds the families of size jjj, and njn_jnj​ is the number of such families. An immigrant founds a new family of size 111 at rate ν\nuν; within a family of size jjj each individual gives birth at rate λ\lambdaλ and dies at rate μ\muμ, so a family moves from colony jjj to colony j+1j+1j+1 at rate jλj\lambdajλ and to colony j−1j-1j−1 at rate jμj\mujμ; a family of size 111 that loses its member becomes extinct and leaves the system at rate μ\muμ. In migration notation this is φj(nj)=nj\varphi_j(n_j) = n_jφj​(nj​)=nj​ with λj,j+1=jλ\lambda_{j,j+1} = j\lambdaλj,j+1​=jλ, λj,j−1=jμ\lambda_{j,j-1} = j\muλj,j−1​=jμ for j≥2j \ge 2j≥2, ν1=ν\nu_1 = \nuν1​=ν and μ1=μ\mu_1 = \muμ1​=μ.

Let 0<λ<μ0 < \lambda < \mu0<λ<μ and ν>0\nu > 0ν>0, and set

αj=νλj(λμ)j,j≥1.\alpha_j = \frac{\nu}{\lambda j}\left(\frac{\lambda}{\mu}\right)^{j}, \qquad j \ge 1 .αj​=λjν​(μλ​)j,j≥1.

Then three things hold.

  1. The traffic equation at colony 111: α1(μ+λ)=ν+2μ α2\alpha_1(\mu + \lambda) = \nu + 2\mu\,\alpha_2α1​(μ+λ)=ν+2μα2​.
  2. The traffic equation at each colony j≥2j \ge 2j≥2:
αj(jλ+jμ)=αj−1(j−1)λ+αj+1(j+1)μ.\alpha_j\bigl(j\lambda + j\mu\bigr) = \alpha_{j-1}(j-1)\lambda + \alpha_{j+1}(j+1)\mu .αj​(jλ+jμ)=αj−1​(j−1)λ+αj+1​(j+1)μ.
  1. The series ∑j≥1αj\sum_{j \ge 1}\alpha_j∑j≥1​αj​ converges, with
∑j≥1αj=−νλlog⁡ ⁣(1−λμ).\sum_{j\ge 1}\alpha_j = -\frac{\nu}{\lambda}\log\!\left(1 - \frac{\lambda}{\mu}\right).j≥1∑​αj​=−λν​log(1−μλ​).

Because φj(n)=n\varphi_j(n) = nφj​(n)=n gives gj=eαjg_j = e^{\alpha_j}gj​=eαj​, the equilibrium numbers of families of each size are independent Poisson variables of means αj\alpha_jαj​, so the third identity is the mean of the total number of distinct families. The condition λ<μ\lambda < \muλ<μ is the stability condition, and it is exactly what makes the series converge.

Formalization Note Colonies are indexed by the positive integers, so the statements are written at j+1j+1j+1, j+2j+2j+2 and j+3j+3j+3 to keep every index positive and avoid truncated subtraction. Convergence of the series and the value of its sum are asserted together.

Preamble
import Mathlib
import Definitions.Def_KellyStochasticNetworks_Migration
Formal statement
namespace KellyStochasticNetworks

theorem family_size_traffic (lam mu nu : ℝ) (hlam : 0 < lam) (hmu : lam < mu) (hnu : 0 < nu)
    (α : ℕ → ℝ)
    (hα : ∀ j : ℕ, α (j + 1) = nu / (lam * ((j : ℝ) + 1)) * (lam / mu) ^ (j + 1)) :
    α 1 * (mu + lam) = nu + α 2 * (2 * mu)
      ∧ (∀ i : ℕ, α (i + 2) * (((i : ℝ) + 2) * lam + ((i : ℝ) + 2) * mu)
            = α (i + 1) * (((i : ℝ) + 1) * lam) + α (i + 3) * (((i : ℝ) + 3) * mu))
      ∧ HasSum (fun j : ℕ => α (j + 1)) (-(nu / lam) * Real.log (1 - lam / mu)) := by sorry

end KellyStochasticNetworks
Source
Kelly & Yudovina, Stochastic Networks, CUP 2014, p. 34 (PDF p. 42), the family-size process of Kendall (1975): rates q(n, T^{j,j+1}n) = j lambda n_j, q(n, T^{j,j-1}n) = j mu n_j, q(n, T^{->1}n) = nu, q(n, T^{1->}n) = mu n_1. 'The traffic equations have a solution (check!) alpha_j = nu/(lambda j) (lambda/mu)^j' and 'the total number of distinct families, N = sum_j n_j, has a Poisson distribution with mean sum_j alpha_j = -(nu/lambda) log(1 - lambda/mu), from the series expansion of log(1-x).' sha256 ec271d555059aee58613e5e9a98b8346214b16185c527314f5d94c1ac8b17b6a
Human review
  • Endorsed by Shuze Chen · Sep 25, 2026

    Confirmed by the moderator at approval.

  • Endorsed by naimengye · Sep 25, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me