Random friend trees

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Berry, Louigi Addario, Briend, Simon, Devroye, Luc, Donderwinkel, Serte, Kerriou, Céline, Lugosi, Gábor
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911819404148736
author Berry, Louigi Addario
Briend, Simon
Devroye, Luc
Donderwinkel, Serte
Kerriou, Céline
Lugosi, Gábor
author_facet Berry, Louigi Addario
Briend, Simon
Devroye, Luc
Donderwinkel, Serte
Kerriou, Céline
Lugosi, Gábor
contents We study a random recursive tree model featuring complete redirection called the random friend tree and introduced by Saramäki and Kaski. Vertices are attached in a sequential manner one by one by selecting an existing target vertex and connecting to one of its neighbours (or friends), chosen uniformly at random. This model has interesting emergent properties, such as a highly skewed degree sequence. In contrast to the preferential attachment model, these emergent phenomena stem from a local rather than a global attachment mechanism. The structure of the resulting tree is also strikingly different from both the preferential attachment tree and the uniform random recursive tree: every edge is incident to a macro-hub of asymptotically linear degree, and with high probability all but at most $n^{9/10}$ vertices in a tree of size $n$ are leaves. We prove various results on the neighbourhood of fixed vertices and edges, and we study macroscopic properties such as the diameter and the degree distribution, providing insights into the overall structure of the tree. We also present a number of open questions on this model and related models.
format Preprint
id arxiv_https___arxiv_org_abs_2403_20185
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Random friend trees
Berry, Louigi Addario
Briend, Simon
Devroye, Luc
Donderwinkel, Serte
Kerriou, Céline
Lugosi, Gábor
Probability
60C05 60J80 05C05
We study a random recursive tree model featuring complete redirection called the random friend tree and introduced by Saramäki and Kaski. Vertices are attached in a sequential manner one by one by selecting an existing target vertex and connecting to one of its neighbours (or friends), chosen uniformly at random. This model has interesting emergent properties, such as a highly skewed degree sequence. In contrast to the preferential attachment model, these emergent phenomena stem from a local rather than a global attachment mechanism. The structure of the resulting tree is also strikingly different from both the preferential attachment tree and the uniform random recursive tree: every edge is incident to a macro-hub of asymptotically linear degree, and with high probability all but at most $n^{9/10}$ vertices in a tree of size $n$ are leaves. We prove various results on the neighbourhood of fixed vertices and edges, and we study macroscopic properties such as the diameter and the degree distribution, providing insights into the overall structure of the tree. We also present a number of open questions on this model and related models.
title Random friend trees
topic Probability
60C05 60J80 05C05
url https://arxiv.org/abs/2403.20185