§5, (50)–(53), p. 11 — for constant rates λ₀ ≤ μ₀, ψ(1, w, ∞) generates the law Q_M of the cumulative population M_∞
ProvedKendallBD.Cumul.cumulative_lawby mikedeng1 · Oct 6, 2026 · Mathlib 0df444a (Lean v4.33.1)
birth-death-processcentral-binomialcumulative-populationgenerating-functionp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Consider a simple birth-and-death process with constant birth rate λ0>0 and death rate μ0≥λ0 (a transient process), started from one ancestor. Let ψ(z,w,t) be the function (50),
ψ(z,w,t)=w(β−z)+(z−α)e−λ0w(β−α)tα(β−z)+β(z−α)e−λ0w(β−α)t,
where α<β are the roots of λ0wz2−(λ0+μ0)z+μ0=0, and let
QM=2λ0λ0+μ022M(M!)2(2M)!2M−1xM(M=1,2,3,…),x=(λ0+μ0)24λ0μ0,
with Q0=0. Then:
- for every w∈(0,1), ψ solves the equation (31), ∂ψ/∂t={λ0wz2−(λ0+μ0)z+μ0}∂ψ/∂z, at every z∈[0,1] and t≥0, and satisfies the boundary condition (32), ψ(z,w,0)=zw for every real z;
- for every w∈(0,1) the series ∑MQMwM converges and
t→∞limψ(1,w,t)=M=1∑∞QMwM;
- QM≥0 for every M and ∑M≥1QM=1.
In words: the joint generating function of the population size and the cumulative population Mt (the number of individuals ever alive) is (50), and the limiting distribution of Mt, read off ψ(1,w,∞), is the proper probability law (QM)M≥1 of (52). For λ0=μ0 the law has an infinite mean; for λ0<μ0 its mean is μ0/(μ0−λ0).
Formalization Note. The process and the joint law Pn,M(t) are not constructed, as in the paper: "the law of M∞" is read, as on p. 11, as the coefficients of the generating function limt→∞ψ(1,w,t), and ψ is tied to the process through the equation (31) and the boundary condition (32). The roots are used only for 0<w<1, where 0<α<1<β; this range determines the coefficients. Q0=0 encodes M∞≥M0=1.
Preamble
import Mathlib
import Definitions.Def_KendallBD_Cumul_Setting
open Filter Topology
Formal statement
namespace KendallBD.Cumul
theorem cumulative_law (lam0 mu0 : ℝ) (hlam : 0 < lam0) (hle : lam0 ≤ mu0) :
(∀ w ∈ Set.Ioo (0:ℝ) 1,
(∀ z ∈ Set.Icc (0:ℝ) 1, ∀ t : ℝ, 0 ≤ t → SolvesPDE lam0 mu0 (psi lam0 mu0) z w t) ∧
∀ z : ℝ, psi lam0 mu0 z w 0 = z * w) ∧
(∀ w ∈ Set.Ioo (0:ℝ) 1,
Summable (fun M : ℕ => Q lam0 mu0 M * w ^ M) ∧
Tendsto (fun t => psi lam0 mu0 1 w t) atTop (𝓝 (∑' M : ℕ, Q lam0 mu0 M * w ^ M))) ∧
((∀ M : ℕ, 0 ≤ Q lam0 mu0 M) ∧ HasSum (Q lam0 mu0) 1) := by sorry
end KendallBD.Cumul
Source
Kendall, On the generalized "birth-and-death" process, Ann. Math. Statist. 19 (1948), §5, (50), (51), (52), (53), p. 11, with §4, (31)–(32), p. 8
View graph
Confirmed by the moderator at approval.