Local convergence of critical random trees and continuous-state branching processes (Q2135185)

This paper is a study of the local convergence of Galton-Watson (GW) trees under a number of different and general classes of conditioning. The first result is about the convergence of a sequence of critical GW trees conditional on monotone functional being larger than \(n\). As \(n\to \infty\), it is shown that a GW tree converges in distribution to the so-called \emph{immortal tree}. Roughly speaking, this is a random tree which has a unique infinite spine and the nodes that are rooted at it have the offspring distribution of the GW tree. The second main theorem of the paper is a similar convergence in distribution if the functional satisfies a certain additivity condition. The convergence regards the GW tree conditional on the functional taking value \(n\) and letting \(n\to \infty\). The limit turns out again to be the immortal tree. Analogous results are obtained for Lévy trees, which are continuum trees that are constructed through Lévy processes.