Scaling limit of multitype Galton-Watson trees with infinitely many types (Q520775)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Scaling limit of multitype Galton-Watson trees with infinitely many types
scientific article

    Statements

    Scaling limit of multitype Galton-Watson trees with infinitely many types (English)
    0 references
    0 references
    0 references
    6 April 2017
    0 references
    Consider a leafed Bienaymé-Galton-Watson (BGW) tree with random edge lengths, i.e., a BGW tree \(T\) with two types, 0 and 1, and random edge length, where only vertices of type 1 can have progeny. Let \(T^1\) be the one-type BGW tree obtained from \(T\) by stripping \(T\) of the ``leaves'', i.e., of all vertices of type 0. Suppose \(T\) has finite first and second moments and \(T^1\) is critical. Denote by \(e(u)\) the type of an edge \(u\), by \(|u|\) its generation, by \(\ell(u)\) its length and assume \(y^2P(\max_{|u|=1,e(u)= 0}\ell(u)> y)\to 0\) and \(y^2E[\sum_{|u|=1,e(u)= 1}1_{\{\ell(u)>y\}}]\to 0\) as \(y\to\infty\). Then, after suitable scaling and normalization, the height of the \(T\) forest converges in distribution to a reflected Brownian motion. Linking the height function of any given multi-type BGW tree to that of a leafed BGW tree with random edge lengths by employing a tree reduction method inspired by \textit{G. Miermont} [Ann. Inst. Henri Poincaré, Probab. Stat. 44, No. 6, 1128--1161 (2008; Zbl 1178.60058)], the convergence result for the height of the \(T\) forest is used to prove, under second moment conditions, an invariance principle for forests of critical BGW trees with a countable set of types. This generalizes Miermont's convergence theorem for forests with a finite set of types. As an application a new approach to a result on random laminations due to \textit{N. Curien} and \textit{Y. Peres} [Electron. Commun. Probab. 16, 435--446 (2011; Zbl 1254.60016)] is provided.
    0 references
    0 references
    Bienaymé-Galton-Watson tree
    0 references
    random edge lengths
    0 references
    scaling limit
    0 references
    random lamination
    0 references
    0 references
    0 references