From sand to networks: a study of multi-disciplinarity

– In this paper, we study empirically co-authorship networks of neighbouring scientific disciplines, and describe the system by two coupled networks. By considering a large time window, we focus on the properties of the interface between the disciplines. We also focus on the time evolution of the co-authorship network, and highlight a rich phenomenology including first order transition and cluster bouncing and merging. Finally, we present a simple Ising-like model (CDIM) that reproduces qualitatively the structuring of the system in homogeneous phases. Introduction. – Since the pioneering works of Barabasi and Albert [1, 2], " complex networks " have become a more and more active field, attracting physicists from the whole sub-fields of statistical physics, ranging from theoretical non-equilibrium statistical physics to experimental granular compaction. These complex structures are usually composed by large number of internal components (the nodes), and describe a wide variety of systems of high technological and intellectual importance, examples including the Internet [3], business relations between companies [4], ecological networks [5] and airplane route networks [6]. As a paradigm for large-scale social networks, people usually consider co-authorship networks [7], namely networks where nodes represent scientists, and where a link is drawn between them if they co-authored a common paper. Their study has been very active recently, due to their complex social structure [8], to the ubiquity of their bipartite structure in complex systems [9] [10], and to the large databases available (arXiv and Science Index). In this paper, we analyze data for such collaboration networks and focus on the development of neighbouring scientific disciplines in the course of time, thereby eyeing the spreading of new ideas in the science community. Let us stress that the identification of the mechanisms responsible for knowledge diffusion and, possibly, scientific avalanches, is primordial in order to understand the scientific response to external political decisions, and to develop efficient policy recommendations. In section 2, we concentrate empirically on this issue by studying data extracted from the arXiv database. To do so, we discriminate two sub-communities of physicists, those studying " complex networks " and those studying " granular media ". This choice is motivated by the relative closeness of these fields, that allows interactions between sub-communities (inter-disciplinarity collaboration), and the passage of a scientist from one field to the other (scientist mobility). The data analysis highlights that most contacts between the two disciplines are driven by inter-disciplinary collaborations, and reveals complex