On a stochastic model for continuous mass branching population
Introduction
The theory of branching stochastic processes is a rapidly developing part of the general theory of stochastic processes. During quite a long time the main object of investigation in the theory of branching processes was the number of individuals (particles) at a given time. So in classic models of branching processes the state-space is the set of non-negative integers. However in many applications one may have situations when it is difficult to count the number of individuals in the population, but some non-negative characteristic, such as volume, weight or product produced by the individuals can be measured. At the end of sixties Jirina [5], [6] defined a branching stochastic process with continuous state space as a homogeneous Markov process the transition probabilities of which satisfy some “branching condition”. Later many papers were published to study this kind of processes (see Refs. [4], [8], [9], [11], [12]). A model of the branching process with continuous state space has appeared in [10] as limiting for branching processes with generalized immigration. Kallenberg [7] introduced a branching model with continuous state-space and studied it under the assumption that the “offspring distribution” is infinitely divisible. Adke and Gadag [1] defined a new model of continuous state-space process with immigration using in the “branching condition” a counting process with independent and stationary increments.
In the paper [1] the authors investigated some distributional properties (and the extinction probability) of the process in the case when the offspring and immigration distributions vary from generation to generation (so-called varying environments). However the asymptotic results are obtained under the assumption of fixed environments. On the other hand in applications the immigration rate may usually be affected by seasonal and global changes of the environment. In this paper we will study asymptotic behavior of the continuous state-space branching stochastic process with immigration in varying environments.
It is convenient to define the process, which we are going to consider, as a family of non-negative random variables describing the amount of a product produced by individuals of the population. The initial state of the process is given by a non-negative random variable . The amount of the product of the first generation is defined as the sum of random products produced by individuals and the product of immigrating to the first generation individuals. Similarly the amount of the product of the second generation is defined as the sum of products produced by individuals and , and so on. Here , are counting processes with independent stationary increments, T is either or and , are non-negative random variables. In the paper [1] limit distributions are obtained for process when are i.i.d. random variables, which corresponds to the fixed environment. In particular when the process is critical it is shown that the linearly normalized process has a gamma limiting distribution. If one does not assume that have a common distribution, several rather important questions appear: (i) under which conditions we may still use the linear normalization to get a non-degenerate limiting distribution? (ii) When will the limiting gamma distribution be preserved? (iii) How does a change of the rate of immigration affect the asymptotic behavior of the process? In this paper we expect to obtain results which give answers for these kinds of questions.
Section snippets
Main theorems
We now give a detailed definition of the process which we are going to consider. Let be a double array of independent and identically distributed non-negative random variables, be a family of non-negative, integer-valued independent processes with independent stationary increments, with almost surely, is either or .
We define a new process as following. Let the initial state of the process be which is an arbitrary non-negative
The Foster–Williamson theorem
As it was indicated before process is a Galton–Watson process with immigration. The offspring distribution and the distribution of the number of immigrating “individuals” have Laplace transforms and , respectively (see [1]). Here and .
We obtain the moments of offspring distribution by standard arguments as following:where and for we obtain:
Acknowledgments
These results are part of the project no. FT-2005/01 funded by KFUPM, Dhahran, Saudi Arabia. I am indebted to King Fahd University of Petroleum and Minerals for excellent research facilities.
References (12)
- S.R. Adke, V.G. Gadag, A new class of branching processes. Branching processes, in: Proceedings of the First World...
An Introduction to Probability Theory and its Applications, vol. 2
(1966)- et al.
Limit theorems for the Galton–Watson process with time-dependent immigration
Z. Wahrschein. Verw. Geb.
(1971) Asymptotic behaviour of continuous time, continuous state-space branching processes
J. Appl. Prob.
(1974)Stochastic branching processes with continuous state-space
Czehost. Math. J.
(1958)Stochastic branching processes with a continuous space of states
Theory Probab. Appl.
(1959)