Critical age-dependent branching Markov processes and their scaling limits (Q601204)

The main objects in the paper are age-dependent branching Markov processes. The distribution of these processes is determined by a triple \((G,(p_k)_{k \geq 0},\eta)\), where \(G\) is a lifetime distribution on \((0,\infty)\) with \(\mu = \int \! s \, G({\mathrm d}s)\), \((p_k)_{k \geq 0}\) is a non-trivial (i.e.\;non-degenerate at 1) offspring distribution on \(\mathbb{N}_0\) and \(\eta = (\eta(t))_t\) is a time-inhomogeneous Markov process starting at \(0\). The Markov branching process \((G,(p_k)_k,\eta)\) is described by its configurations \(\mathcal{C}_t\) at times \(t \geq 0\), where \[ \mathcal{C}_t = \{(a_t^i,X_t^i): i = 1,\ldots, N_t\} \] with \(N_t\) being the number of particles present at time \(t\), \(a_t^i\) describing the age of the \(i\)th particle alive at time \(t\), and \(X_t^i\) denoting the corresponding position. A particle born at time \(\tau\) at \(x \in \mathbb{R}\) moves in \(\mathbb{R}\) according to the process \(\{x + \eta(t-\tau): \tau \leq t < \tau + L\}\) where \(L\) has distribution \(G\). At time \(\tau+L\) and position \(x + \eta(L)\), the particle dies and produces offspring according to a random variable \(\xi\) with distribution \((p_k)_k\). The three objects \(L\), \(\xi\), and \(\eta\) associated with each particle are assumed to be independent, the tuples \((L,\xi,\eta)\) for different particles are assumed to be i.i.d. The authors consider the situation when the second moments of \(L\) and \(\xi\) are finite and \(\eta\) is centered and has locally uniformly bounded second moment function. Further, they assume \(\int \mathbb{E} \eta(s)^2 \, G({\mathrm d}s) < \infty\) and criticality of the offspring distribution, that is, \(\sum_{k \geq 1} kp_k = 1\). In this situation, it is shown that given \(\{N_t > 0\}\), with \(a_t\) and \(X_t\) denoting age and position of a particle chosen uniformly at random from the particles present at time \(t\), \((a_t,X_t/\sqrt{t})\) converges in distribution to a tuple \((U,V)\) of independent random variables with \(\mathbb{P}(U \in {\mathrm d}x) = \mu^{-1}(1-G(x))1_{(0,\infty)}(x){\mathrm d}x\) and \(V\) having standard normal distribution. Further, with \[ Y_t := \sum_{i=1}^{N_t}\delta_{(a_t^i,X_t^i)} \quad \text{and} \quad \tilde{Y}_t(A\times B) := Y_t(A \times \sqrt{t}B), \quad A \subseteq [0,\infty),\;B \subseteq \mathbb{R}, \] the convergence in distribution of \(\tilde{Y}_t/N_t\) given \(\{N_t>0\}\) is examined. Finally, the weak convergence of a sequence of age-dependent critical Markov branching processes on the Skorokhod space is investigated.