The local limit theorem for supercritical branching processes with immigration (Q6046192)

Consider \((X_n, n \geq 0)\) a supercritical Galton-Watson branching process with immigration, initiated by a random number \(X_0\) of initial individuals. In this process, at every generations each individual creates independently of one another a random number of children according to the probability distribution \((p_n)\) satisfying \(p_0=0\). Additionally, an independent number of immigrants, sampled according to the probability distribution \((h_n)\), are added to the population. Let \(Z_n\) be the number of individuals alive at time \(n\) and \((c_n)\) be a real-valued sequence such that \(E(e^{-Z_n/c_n}) = e^{-s_0}\) for some fixed \(s_0>0\). \textit{E. Seneta} proved that \((X_n/c_n)\) converges almost surely to a random variable \(V\) [Math. Biosci. 7, 9--14 (1970; Zbl 0209.48804)], implying in particular its convergence in distribution. In the present article, the authors prove a local limit theorem, by determining the limit of \(c_n P(X_n =k_n)\) as \(n \to \infty\) for any sequence \((k_n)\) such that \(k_n/c_n \to t\).