scientific article; zbMATH DE number 3856890

From MaRDI portal
Publication:3325531





A Markov branching process \(\xi\) (t) has proliferation intensities \(\{\pi_ m(k)\}\), \(m\geq 0\), if k particles are present. \(\pi_ m(k)=\pi_ m\) for \(k\leq N\), while \(\pi_ m(k)=N\pi_ m/k\) for \(k>N\), where N represents the environment limitations \((\sum_{m}\pi_ m=0)\). [Remark: In the paper, the process definition is self-contradictory, due to a printing error. The above definition is based on the author, ibid. 34, 681-687 (1982; Zbl 0508.60069)]. Immigration is allowed; in a small interval \((t,t+\Delta t)\), m new particles arrive with probability \(\delta_{0m}+\epsilon_ m\Delta t+o(\Delta t) (\sum_{m}\epsilon_ m=0).\) The author considers the life periods of the process, defined as time intervals \([T,T+\tau)\), with \(\xi(t)>0\), \(t\in [T,T+\tau)\) and \(\xi(T- 0)=\xi(T+\tau)=0\). The method is in principle, based on writing down the Kolmogorov equations, solving them in the Laplace transform domain and using a Tauberian theorem to recover the properties in the time domain [see also the author's paper in: Some questions of the theory of stochastic processes, Coll. sci. Work, Kiev 1982, 3-10 (1982; Zbl 0531.60080)]. On the life periods, the process behaves differently, depending on whether \(N\phi '(1)+\omega '(1)\) is (a) negative, (b) zero or (c) positive, where \(\phi\) (z) and \(\omega\) (z) are generating functions of sequences \((\pi_ m)\) and \((\epsilon_ m)\), respectively. In the case (a) \(P(\tau<\infty)=1\), \(E(\tau)<\infty\); in the case (b) \(P(\tau>t)\sim t^{-1/2}\), for t large; in the case (c) \(P(\tau<\infty)=\alpha \leq 1\), \(E(\tau)=\infty\). Also, results are proved for the distribution of \(\eta\) (t) (i.e. \(\xi\) (t) restricted to the life period and shifted back by T), conditioned on \(\{\tau>t\}\).











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3325531)