Preferential attachment combined with random number of choices

TL;DR

This paper analyzes how the maximum degree in an evolving tree model behaves asymptotically, considering the effects of local choices and preferential attachment with random choices, revealing different growth regimes based on the expected number of choices.

Contribution

It provides a rigorous analysis of the maximal degree behavior in a combined preferential attachment and local choice model, including phase transitions and asymptotic distributions.

Findings

Maximal degree is sublinear if expected choices < 2+beta

Maximal degree is linear if expected choices > 2+beta

Maximal degree is of order n/ln n when expected choices equals 2+beta

Abstract

We study an asymptotical behavior of the maximal degree in the degree distribution in an evolving tree model combining the local choice and the Mori's preferential attachment. In the considered model, the random graph is constructed in the following way. At each step, a new vertex is introduced. Then, we connect it with one (the vertex with the largest degree is chosen) of $d$ (for random $d$) possible neighbors, which are sampled from the set of the existing vertices with the probability proportional to their degrees plus some parameter $\beta>-1$. It is known that the maximum of the degree distribution for non-random $d>2$ has linear behaviour and, for $d=2$, asymptoticaly equals to $n/\ln n$ up to a constant factor. We prove that if $\mathbb{E}d<2+\beta$, the maximal degree has sublinear behavior with the power $\mathbb{E}d/(2+\beta)$ (as in the preferential attachment without choice), if $\mathbb{E}d>2+\beta$, it has linear behavior and if $\mathbb{E}d=2+\beta$ the maximal degree is of order $n/\ln n$. The proof combines standard preferential attachment approaches with martingales and stochastic approximation techniques. We also use stochastic approximation results to get a multidimentional central limit theorem for the numbers of vertices with fixed degrees.