CONICET | Buscador de Institutos y Recursos Humanos

Abstract. Activity-driven modelling has recently been proposed as analternative growth mechanism for time varying networks,displaying power-lawdegree distribution in time-aggregated representation. This approach assumesmemoryless agents developing random connections with total disregard of theirprevious contacts. Thus, such an assumption leads to time-aggregated randomnetworks that do not reproduce the positive degree-degree correlation and highclustering coefficient widely observed in real social networks. In this paper, weaim to study the incidence of the agents? long-term memory on the emergence ofnew social ties. To this end, we propose a dynamical network model assumingheterogeneous activity for agents, together with a triadic-closure step as mainconnectivity mechanism. We show that this simple mechanism provides some ofthe fundamental topological features expected for real social networks in theirtime-aggregated picture. We derive analytical results and perform extensivenumerical simulations in regimes with and without population growth. Finally,we present an illustrative comparison with two case studies, one comprising faceto-face encounters in a closed gathering, while the other one corresponding tosocial friendship ties from an online social network