Weighted height of random trees.pdfVIP

Weighted height of random trees N. Broutin L. Devroye E. McLeish? January 3, 2008 Abstract We consider a model of random trees similar to the split trees of Devroye [30] in which a set of items is recursively partitioned. Our model allows for more flexibility in the choice of the partitioning procedure, and has weighted edges. We prove that for this model, the height Hn of a random tree is asymptotic to c log n in probability for a constant c that is uniquely characterized in terms of multivariate large deviations rate functions. This extension permits us to obtain the height of pebbled tries, pebbled ternary search tries, d-ary pyramids, and to study geometric properties of partitions generated by k-d trees. The model also includes all polynomial families of increasing trees recently studied by Broutin, Devroye, McLeish, and de la Salle [17]. Keywords and phrases: Height, random tree, split tree, branching process, large deviations, probabilistic analysis. 1 Introduction In this paper, we propose a general framework for finding heights of random trees with bounded degree d using branching processes. This extends previous work on the heights of a handful of random tree models [11, 14, 17, 27, 30]. Our model makes the connection between two different ways of constructing random trees: the split trees of Devroye [30] based on the partition of a set of items into recursive bins, and the branching process techniques of Broutin and Devroye [14] and generalizes both of them. The resulting framework encompasses all the kinds of trees that were captured by other branching process techniques such as binary search trees [27], random recursive trees [28, 56], plane oriented trees [56], median-of-(2k + 1) trees [66], d-ary pyramids [11] and increasing trees [17]. The split tree model is a very natural way to build a random tree: a tree shall be seen as the outcome of a recursive process partitioning a finite set of items. ?Research of the authors was supported by NSERC Grant A3