A stochastic discrete generation birth, continuous death population growth model and its approximate solution (Q799533)

The authors consider a stochastic generalisation of a population model first introduced by \textit{E. T. Poulsen}, ibid. 8, 325-343 (1979; Zbl 0442.92020). Birth events occur at times \(t_ n\) where \(\tau_ n=t_ n- t_{n-1}\) (\(n\geq 1\), \(t_ 0=0)\) are independent and identically distributed random variables. Let X(t) denote the population size at time t, \(X_ n=X(t_ n-)\) and \(Z_ n=X(t_ n+).\) Given \(Z_ n\), the population changes deterministically during \((t_ n,t_{n+1})\) according to the differential equation \(X'(t)=-r(X(t))\) where \(r(0-)=0\) and \(0<r(x)<\infty\) if \(0<x\leq N\leq\infty \). Given \(X_ n=x_ n\) the population size just after the birth event at \(t_ n\) is a random variable \(S_ n(x_ n)\) and \(P(S_ n(x_ n)\leq y)=\int^{y}_{0}b(u,x_ n)du\). The authors pay particular attention to the cases where the \(\tau_ n\) are equal and constant and \(r(\cdot)\) corresponds to logistic or Gompertzian growth. The sequence \(\{Z_ n\}\) satisfies a first order non-linear stochastic difference equation and, apart for a couple of special cases, it is impossible to find tractable expressions for the distributions of the \(Z_ n\), or even the limiting distribution if it exists. They show how to obtain good approximations by using Gram-Charlier expansions and illustrate their methodology with several numerical examples. Comparisons are also made with some specific Ito stochastic differential equation models for population growth in random environments.