Kosto V. Mitov,Nikolay M. Yanev
Abstract:The paper studies a class of critical Markov branching processes with infinite variance of the offspring distribution. The processes admit also an immigration component at the jump-points of a non-homogeneous Poisson process, assuming that the mean number of immigrants is infinite and the intensity of the Poisson process converges to a constant. The asymptotic behavior of the probability for non-visiting zero is obtained. Proper limit distributions are proved, under suitable normalization of the sample paths, depending on the offspring distribution and the distribution of the immigrants.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to study a class of critical Markov branching processes with non - homogeneous Poisson immigration (MBPNPI), in which the offspring distribution has an infinite variance. Specifically, the author focuses on the following aspects:
1. **Model description**:
- The research object is a single - type Markov branching process \(Z(t)\), in which each particle evolves independently, the lifetime follows an exponential distribution, and a random number of new particles are generated at the end of the life.
- Immigrants generated by a non - homogeneous Poisson process are also introduced into the process. These immigrants enter the system at the jump points of the Poisson process and participate in the subsequent evolution.
2. **Key assumptions**:
- The form of the offspring probability generating function (p.g.f.) \(f(s)\) is:
\[
f(s)=s+(1 - s)^{\gamma + 1}L\left(\frac{1}{1 - s}\right),\quad s\in[0,1)
\]
where \(\gamma\in(0,1]\), and \(L(\cdot)\) is a slowly varying function (s.v.f.) at infinity.
- The form of the immigrant p.g.f. \(g(s)\) is:
\[
g(s)=1-(1 - s)^{\alpha}l\left(\frac{1}{1 - s}\right),\quad s\in(0,1)
\]
where \(\alpha\in(0,1]\), and \(l(x)\) is also a slowly varying function at infinity.
3. **Main problems**:
- **Asymptotic behavior of non - zero visit probability**: Study the asymptotic behavior of the probability that the process \(Y(t)\) does not visit the zero state for a long time.
- **Limit distributions**: Under appropriate normalization conditions, explore the limit distributions of the process \(Y(t)\), which are specifically divided into four cases:
1. When \(tq(t)\to\infty\), the limit distribution is a one - dimensional stable distribution with parameter \(\alpha\).
2. When \(tq(t)\to C\in(0,\infty)\), the limit distribution belongs to the normal attraction domain of the stable law with parameter \(\gamma\).
3. When \(Q = \int_{0}^{\infty}q(t)dt<\infty\), the conditional limit distribution is discrete and has an infinite mean.
4. When \(tq(t)\to0\) but \(\int_{0}^{\infty}q(t)dt=\infty\), the limit distribution is uniformly distributed on the unit interval.
Through these studies, the author aims to reveal the influence of the immigration and offspring distribution characteristics on the long - term behavior of the system in such branching processes, especially in the case of an infinite offspring variance. This is helpful for understanding practical problems such as cell dynamics in biological systems.