Abstract
We study branching Markov chains on a countable state space (space of types) with the focus on the qualitative aspects of the limit behaviour of the evolving empirical population distributions. No conditions are imposed on the multitype offspring distributions at the points of other than to have the same average and to satisfy a uniform moment condition. We show that the arising population martingale is uniformly integrable. Convergence of population averages of the branching chain is then put in connection with stationary spaces of the associated ordinary Markov chain on (assumed to be irreducible and transient). Our principal result is the almost sure convergence of the empirical distributions to a random probability measure on the boundary of an appropriate compactification of . Final considerations concern the general interplay between the measure theoretic boundaries of the branching chain and the associated ordinary chain.
Nous étudions les chaînes de Markov branchantes sur un espace d’états (espace de types) dénombrable en mettant l’accent sur les aspects qualitatifs du comportement limite de l’évolution des distributions empiriques de la population. Aucune condition n’est imposée sur les distributions multitypes des descendants des points de autre que d’avoir la même moyenne et de satisfaire à une condition de moment de type . Nous montrons que la martingale de population résultante est uniformément intégrable. Ensuite, nous établissons le lien entre la convergence des moyennes empiriques de la chaîne branchante et les espaces stationnaires de la chaîne de Markov ordinaire associée sur (supposée irréductible et transiente). Notre résultat principal est la convergence presque sûre des distributions empiriques vers une mesure de probabilité aléatoire sur le bord d’une compactification appropriée de . Les considérations finales portent sur l’interaction générale entre les bords mesurables de la chaîne branchante et de la chaîne ordinaire associée.
Funding Statement
The second author was supported by Austrian Science Fund project FWF: P31889-N35 during a visit at University of Ottawa in 2019.
Acknowledgements
The beginning of this work has its roots in a discussion of the authors with Elisabetta Candellero in Warwick in 2015, where the first author of the present paper proposed to study the behaviour of the sequence of empirical distributions rather than of individual genealogical lines. Although a recent preprint of CANDELLERO–HUTCHCROFT [14] (which was posted on arXiv virtually simultaneously with the present article) also stems from that discussion, the approach there is quite different. The authors consider branching Markov chains with independent branching and displacement (see Example 2.13) and establish the boundary convergence of the rescaled empirical distributions with respect to the Martin compactification of the underlying chain. It might be tempting to look at the possible ramifications of a combination of these two approaches.
Citation
Vadim A. Kaimanovich. Wolfgang Woess. "Limit distributions of branching Markov chains." Ann. Inst. H. Poincaré Probab. Statist. 59 (4) 1951 - 1983, November 2023. https://doi.org/10.1214/22-AIHP1344
Information