Title:Network tomography: a new structure spectrum for studying network dynamics

Abstract:The analysis of dynamics on complex networks is closely connected to structural features of the network itself. Network features like, for instance graph-cores and vertex degrees have been proven to be discriminants of network dynamics and are studied ubiquitously. Here we connect these central discriminants of network dynamics by means of the D-spectrum, a novel framework that identifies a collection of nested chains of subgraphs contained in the network. Cores and vertex degrees can be derived from two particular such chains. Furthermore, each chain gives rise to a rank of a vertex and the collection of these ranks is the D-spectrum of a vertex, providing a novel characterization. We compute this D-spectrum via a node deletion algorithm and we introduce a particular graph dynamical system, called M-C system, in which the D-spectrum appears as a fixed point. Using the D-spectrum we identify nodes of similar spreading power in the susceptible-infectious-recovered (SIR) model on various networks and show that it is more reliable than that based on $k$-core decomposition. In addition we compare the two clusterings obtained by means of vertex infectiousness for different infection probabilities and the D-spectrum and show that they are highly correlated.