Critical non-local spatial branching processes with infinite variance conditioned on survival

TL;DR

This paper studies critical non-local branching processes with infinite variance, establishing conditions for survival probability decay and distributional stability conditioned on survival, extending classical results to more general non-local superprocesses.

Contribution

It provides new sufficient conditions for survival probability decay and Yaglom-type limits in non-local branching processes with infinite variance, generalizing previous spatial and non-spatial models.

Findings

Polynomial decay of survival probability at criticality

Yaglom limit for conditioned process distribution

Framework applicable to non-local superprocesses

Abstract

We consider the setting of either a general non-local branching particle process or a general non-local superprocess. Under the assumption that the mean semigroup has a Perron-Frobenious type behaviour in combination with a regularly varying assumption on the reproductive point process, which permits infinite second moments, we consider sufficient conditions that ensure limiting distributional stability when conditioned on survival at criticality. We offer two main results. Under the aforesaid conditions, our first main contribution establishes the polynomial decay in time of the survival probability in the spirit of a classical Kolmogorov limit. The second main contribution pertains to the stability, when conditioning on survival, in the spirit of a Yaglom limit. In both cases our proofs work equally well for the analogous setting of non-local superprocesses with non-local branching. Our results complete a series of articles for various families of non spatial branching processes from the 1950s-1970s as well complementing a recent result of this type for spatial branching processes of Ren, Song and Sun (2020) The generality of our results improve on older work of Hering and Hoppe (1981) and Asmussen and Hering (1983) who dealt with a similar context for branching particle systems, as well as providing a general framework for non-local superprocesses.