On aggregation of subcritical Galton-Watson branching processes with regularly varying immigration (Q831317)

Let \((X_k)_{k \in \mathbb{N}_0}\) be a Galton-Watson process with i.\,i.\,d.\ immigration and with the immigration process independent of all offspring variables. Let \(\xi\) and \(\varepsilon\) denote generic random variables with \(\xi\) and \(\varepsilon\) having as distribution the offspring law and the immigration law, respectively. The authors assume that \(m_\xi = \mathbb{E}[\xi] \in [0,1)\) and \(\mathbb{E}[\log_+\varepsilon] < \infty\), in which case the Markov chain \((X_k)_{k \in \mathbb{N}_0}\) admits a unique stationary distribution \(\pi\). They further assume that \(X_0\sim\pi\), which guarantees that \((X_k)_{k \in \mathbb{N_0}}\) is a stationary process, and that the tail probability \(\mathbb{P}(\varepsilon > x)\) is regularly varying with index \(\alpha \in (0,2)\) as \(x \to \infty\), in particular, \(\varepsilon\) has infinite second moment. If \(\alpha \in [1,2)\), the authors additionally suppose \(\mathbb{E}[\xi^2] < \infty\). It is known that then \(\mathbb{P}(X_0 > x)\) is also regularly varying with index \(\alpha \in (0,2)\) as \(x \to \infty\), in particular, there exist scaling constants \(a_N > 0\) such that \(N \mathbb{P}(X_0 > a_N) \to 1\) as \(N \to \infty\). The authors consider a sequence of independent copies \((X_k^{(j)})_{k \in \mathbb{N}_0}\), \(j \in \mathbb{N}\) of \((X_k)_{k \in \mathbb{N}_0}\) and prove that \[ \frac1{a_N} \bigg(\sum_{j=1}^{\lfloor Nt \rfloor} (X_0^{(j)} - \mathbb{E}[X_0^{(j)} \mathbf{1}_{\{X_0^{(j)} \leq a_N\}}]), \ldots, \sum_{j=1}^{\lfloor Nt \rfloor} (X_k^{(j)} - \mathbb{E}[X_k^{(j)} \mathbf{1}_{\{X_k^{(j)} \leq a_N\}}]) \bigg)_{t \geq 0} \] converges in distribution in the Skorokhod space to a \((k+1)\)-dimensional \(\alpha\)-stable process with explicitly known characteristic function at time \(1\). The authors also present limit theorems for the aggregated stochastic process \((\sum_{k=1}^{\langle nt \rangle} \sum_{j=1}^N X_k^{(j)})_{t \geq 0}\) with different shift and scaling constants and in an iterated manner such that first \(N\) and then \(n\) converge to infinity.