Limit theorems for a strongly supercritical branching process with immigration in random environment (Q2138600)

The author considers a strongly supercritical branching process with immigration in a random environment (BPIRE), stopped at (large) time \(n\). Let \(\xi^{(n)}_{i}\) be all i.i.d. according to \(F_{n+1}\), and \(\eta_{n}\sim G_{n+1}\). The \(n\)-th generation of the BPIRE is the process \[ Z_{n} = \sum\limits_{i=1}^{Z_{n+1}+\eta_{n}}\xi_{i}^{(n)}. \] The random variables \(\xi^{(n)}_{i}\) are the offspring, and \(\eta_{n}\) is the number of immigrants. The random environment is \((F_{n},G_{n})\), which are random elements from a particular space of probability measures on \(\{0,1,\ldots,\}\). Let \(f_{n}\) and \(g_{n}\) be the generating functions of \(F_{n}\) and \(G_{n}\) respectively. The author considers those distributions whose generating functions are of the form \(f_{n}(s) = p_{n}/(1-(1-p_{n})s)\) and \(g_{n}(s)=s\). The \textit{associated random walk} is the process \[ S_{0}=0, \quad S_{n}=\sum\limits_{i=1}^{n}\frac{1-p_{i}}{p_{i}}. \] The BPIRE is strongly supercritical, i.e., \[ \mathrm{E}\left[\frac{1-p_{i}}{p_{i}}\right] > 0, \text{ and } 0<\mathrm{E}\left[(\frac{1-p_{i}}{p_{i}})e^{\frac{1-p_{i}}{p_{i}}}\right]<\infty. \] Denote the extinction time of the process by \(T^{(n)}=\min\{i\ge n: Z_{i}^{(n)}=0 \}\). First, the author derives a formula for the multidimensional generating function of the process, as the inverse of a sum of elements calculated from the exponential of the previously mentioned associated random walk, i.e., \(e^{S_{n}}\). Using this as the basic tool, the author finds the following. \begin{itemize} \item[1.] The process \(\{Z^{(n)}_{\lfloor nt \rfloor}, t\in [0,1] \vert T^{(n)} < \infty \}\) converges in the sense of finite-dimensional distributions to the process \(\{\zeta(t),t\in[0,1] \}\), where the limiting process is a non-negative integer random process, started at \(0\), where all \(t\in (0,1]\) coordinates are independent. Also the generating function of \(\zeta(1)\) is found. \item[2.] The process \(\{ Z^{(n)}_{n+k}, t\in [0,1] \vert T^{(n)} < \infty \}\) converges in distribution to \(\{\zeta_{k} \}\), where \(\zeta_{k}\) are non-negative integer random variables that are non-zero with positive probability. \item[3.] The extinction time \(\{T^{(n)}-n \vert T^{(n)} < \infty \}\) converges in distribution to a non-negative integer variable, that is finite with probability \(1\). \end{itemize}