Multi-scale clustering for a non-Markovian spatial branching process (Q2576798)

The authors consider a branching diffusion in \(R^d\) constructed from a symmetric \(\alpha\)-stable motion, i.e. a time-homogeneous Markov process with generator \(-(-\Delta)^{-\alpha/2}\), \(\alpha\in (0,2]\), a position-independent, but not necessarily exponential particle-lifetime with finite first moment, and an offspring law of index \(1 +\beta\), i.e., with generating function \(s + c (1-s)^{1+\beta}\), \(c\in (1, 1/(1+\beta)]\). The particles move and branch independently, and the branching is local. When \(d=\alpha/\beta\), multi-scale clustering occurs. With an increasing localized population or increasing homogeneous Poisson population as initial state a corresponding fdd scaling limit is obtained. The limit state is uniform, its intensity varying in line with the scaling index according to a continuous state branching process of index \(1+\beta\). As to the result, this generalizes the theory known for processes with exponential life-time and Brownian motion, but a different method of proof had to be found: The classical PDE method does not apply, instead the asymptotic properties of scaled solutions of the integral equation of \textit{I. Kaj} and \textit{S. Sagitov} [J. Theor. Probab. 11, No.~1, 225--257 (1998; Zbl 0909.60081)] come into play.