Networks with preferred degree: A mini-review and some new results

TL;DR

This paper reviews adaptive social network models with preferred degrees, presents new exact results for the XIE model, and explores phase transitions and degree distribution variations in these systems.

Contribution

It provides the first exact solution for the XIE model, analyzes phase transitions, and introduces variants with preferential attachment, expanding understanding of adaptive network dynamics.

Findings

Exact distribution and Hamiltonian for the XIE model.

Identification of a phase transition with an extreme Thouless effect.

Altered degree distributions in models with preferential attachment.

Abstract

Since their inception about a decade ago, dynamic networks which adapt to the state of the nodes have attracted much attention. One simple case of such an adaptive dynamics is a model of social networks in which individuals are typically comfortable with a certain number of contacts, i.e., preferred degrees. This paper is partly a review of earlier work of single homogeneous systems and ones with two interacting networks, and partly a presentation of some new results. In general, the dynamics does not obey detailed balance and the stationary distributions are not known analytically. A particular limit of the latter is a system of extreme introverts and extroverts - the XIE model. Remarkably, in this case, the detailed balance condition is satisfied, the exact distribution and an effective Hamiltonian can be found explicitly. Further, the model exhibits a phase transition in which the total number of links in the system - a macroscopically interesting quantity, displays an extreme Thouless effect. We show that in the limit of large populations and away from the transition, the model reduces to one with non-interacting agents of the majority subgroup. We determine the nature of fluctuations near the transition. We also introduce variants of the model where the agents show preferential attachment or detachment. There are significant changes to the degree distributions in the steady state, some of which can be understood by theoretical arguments and some remain to be explored. Many intriguing questions are posed, providing some food for thought and avenues for future research.