Representation and approximation of large population age distributions using Poisson random measures (Q1115372): Difference between revisions

Investigated are Wang's population processes [\textit{F. J. S. Wang}, J. Math. Biol. 2, 373-400 (1975; Zbl 0324.60064)] defined for all \(n>0\). The population size process is denoted by \(Z_ n(t)\), the associated population density process is \(P_ n(t)=Z_ n(t)/n.\) (i) The process starts with \(Z_ n(0)=Z\) individuals with age structure \(S(n)=\{S_ 1,...,S_ z\}\), that is, \(0\leq S_ 1\leq...\leq S_ z\) represent the ages of Z initial individuals at time \(t=0\). (ii) The lifetimes of new individuals are i.i.d. with distribution function \(1- F(t)=\mu ([0,t])\), where \(\mu\) is a probability measure on [0,\(\infty)\). (iii) The probability that a particular individual of age s gives birth to a brood of size \(k\geq 1\), during a small time interval \([t,t+\Delta t]\) is \(\alpha_ k(s,P_ n(t))\Delta t+o(\Delta t)\) where the functions \(\alpha_ k(s,x)\) are right continuous in s and \(\sum^{\infty}_{k=1}k\sup_{s,x}\alpha_ k(s,x)<\infty.\) For this process and its extensions, the original results of \textit{T. G. Kurtz} [see: Approximation of population processes, CBMS-NSF Reg. Conf. Ser. Appl. Math. 36 (1981; Zbl 0465.60078)] and Wang are generalized. By a random clock argument it is possible to relate the population process to a countable set of Poisson processes. The deterministic age distribution equation for these Poisson processes approximates the random equations for the original process. Then, the fluctuations of the age structure around its limiting distribution are considered.