Asymptotic properties of Bellman-Harris critical branching processes starting with a large number of particles (Q1175843)

Let \(z(t,B)\) be a critical Bellman-Harris branching process with an initial number of \(B\) particles. Suppose that the probability generating function \(f\) of the number of offsprings, and the distribution function \(G\) of the particle lifetime satisfy: (1) \(f(s)=s+(1-s)^{1+\alpha}L(1-s)\), \(0<\alpha\leq 1\), \(L(x)\) slowly varying as \(x\to 0+\), (2) \(1-G(t)\sim t^{-\beta}l(t)\), \(t\to\infty\), where \(\beta\geq 1\), \(l(t)\) slowly varying for \(t\to\infty\), (3) \(\lim_{t\to\infty}t(1-G(t))/(1-f_ t(0))=\infty\). Let \(R(x)=x^ \alpha L(x)\), \(U(t)=\sum_{k=0}^ \infty G^{*k}(t)\), and assume that the initial number of particles \(B=B(t)\to\infty\) as \(t\to\infty\) so that \(R(a/B(t))\sim(\alpha U(t))^{-1}\), \(a>0\). The author proves that \(\psi(t)z(t,B(t))\) converges in distribution as \(t\to\infty\) under suitable conditions on the normalizing function \(\psi(t)\). The Laplace transform of the limiting distribution is identified.