Branching process: Why does the population die or explode?

119 Views Asked by At

Consider a population such that each member, independently from other members, at a certain instant of time is replaced by its offspring. Lets denote with $X_n$ $({n\ge 1})$ the amount of the population at the the $n$-th generation, and let $X_0$ be the starting population. If $$T_0=\inf\{n\in\mathbb N\,:\, X_n=0\},\quad(\inf({\emptyset})=+\infty)$$ The probability that the population dies, given $X_0=1$, is: $$\alpha:=P(T_0<\infty\,|\,X_0=1)$$


Now reading the book "Bosq, Nguyen - A Course in Stochastic", I don't understand a passage at page 71:

Since the subset of states $\{1,2,\ldots\}$ is a transient class (I understand this), we see that: $$\lim_{n\to\infty} X_n(\omega) =\left\{\begin{array} {ll} 0 & \textrm{if}\; T_0(\omega)<\infty\\ \infty & \textrm{if}\; T_0(\omega)=\infty\\ \end{array}\right.$$

Since $0$ is an absorbing state, the first line of the equation is clear, but why if the population does not die, then it explodes? Why an oscillating behavior can't occur?

Thanks in advance.

1

There are 1 best solutions below

2
On

There can be no oscillations, because then there would be a state that you visit infinitely many times. Each time you visit that state, there is a positive probability for the process to go extinct. For example, just the probability of having no children in the next generation. Denote this probability $p > 0$. Since you visit this state infinitely many times, the probability to $not$ go extinct will be $(1-p)^{\infty} = 0.$