Asymptotic theory of time varying networks with burstiness and heterogeneous activation patterns. (17th May 2017)

• Record Type: Journal Article
• Authors: Burioni, Raffaella; Ubaldi, Enrico; Vezzani, Alessandro
• Is Part Of: Journal of statistical mechanics. (2017:May)
• Publication Date: 2017-05-17
• Subjects: 11 -- 12; Statistical mechanics -- Periodicals; Mechanics -- Statistical methods -- Periodicals; 530.1305
• DOI: 10.1088/1742-5468/aa6ce7
• Languages: English
• ISSNs: 1742-5468

Abstract

Abstract: The recent availability of large-scale, time-resolved and high quality digital datasets has allowed for a deeper understanding of the structure and properties of many real-world networks. The empirical evidence of a temporal dimension prompted the switch of paradigm from a static representation of networks to a time varying one. In this work we briefly review the framework of time-varying-networks in real world social systems, especially focusing on the activity-driven paradigm. We develop a framework that allows for the encoding of three generative mechanisms that seem to play a central role in the social networks' evolution: the individual's propensity to engage in social interactions, its strategy in allocate these interactions among its alters and the burstiness of interactions amongst social actors. The functional forms and probability distributions encoding these mechanisms are typically data driven. A natural question arises if different classes of strategies and burstiness distributions, with different local scale behavior and analogous asymptotics can lead to the same long time and large scale structure of the evolving networks. We consider the problem in its full generality, by investigating and solving the system dynamics in the asymptotic limit, for general classes of ties allocation mechanisms and waiting time probability distributions. We show that the asymptotic network evolution is driven by a few characteristics of these functional forms, that can be extracted from direct measurements on large datasets.