Regenerative tree growth: binary self-similar continuum random trees and Poisson-Dirichlet compositions

From MaRDI portal
(Redirected from Publication:1035868)



Abstract: We use a natural ordered extension of the Chinese Restaurant Process to grow a two-parameter family of binary self-similar continuum fragmentation trees. We provide an explicit embedding of Ford's sequence of alpha model trees in the continuum tree which we identified in a previous article as a distributional scaling limit of Ford's trees. In general, the Markov branching trees induced by the two-parameter growth rule are not sampling consistent, so the existence of compact limiting trees cannot be deduced from previous work on the sampling consistent case. We develop here a new approach to establish such limits, based on regenerative interval partitions and the urn-model description of sampling from Dirichlet random distributions.


Consider a stochastic binary tree growth process \(\{T_n; n\geq 1\}\), constructed according to the following \((\alpha,\theta)\)-selection rule: Let \(0\leq\alpha\leq 1\) and \(\theta\geq 0\). (i) For \(n\geq 2\), the tree \(T_n\) branches at the branch point adjacent to the root into two sub-trees \(T_{n,0}\) and \(T_{n,1}\). Given these are of sizes \(m\) and \(n-m\), say, where \(T_{n,1}\) contains the smallest label in \(T_n\), assign the weight a to the edge connecting the root and the adjacent branch point, weights \(m-\alpha\) and \(n-m-1+\theta\), respectively, to the sub-trees. (ii) Select the root edge or a sub-tree with probabilities proportional to these weights. If a sub-tree with two or more leaves was selected, recursively apply the weighting procedure (i) to the selected sub-tree, until the root edge or a sub-tree with a single leaf was selected. If a sub-tree with a single leaf was selected, select the unique edge of this sub-tree. The limit theory of \textit{B. Haas}, \textit{G. Miermont}, \textit{J. Pitman} and \textit{M. Winkel} [Ann. Probab. 36, No. 5, 1790--1837 (2008; Zbl 1155.92033)], covers the special case \(\alpha+\theta= 1\), but relies on sampling consistency: (Let \(T^0_n\) be obtained from \(T_n\) by removing the leaf labels, and \(T^0_n\) from \(T^0_{n-1}\) by removing a leaf chosen uniformly at random. \(\{T_n; n\geq 1\}\) is called weakly sampling consistent if the distributions of \(T^0_n\) and \(\widehat T^0_n\) coincide for all \(n\geq 1\).) In general, \((\alpha,\theta)\)-tree growth processes are not weakly sampling consistent. So, the authors now take a new approach to the existence of compact limiting trees, which applies to the general case. It is based on regenerative interval partitions and the urn-model description of sampling from Dirichlet random distributions.



Cites work


Cited in
(23)








This page was built for publication: Regenerative tree growth: binary self-similar continuum random trees and Poisson-Dirichlet compositions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1035868)