Title:Propagation on networks: an exact alternative perspective

Abstract:By generating the specifics of a network structure only when needed (on the fly), we derive a simple stochastic process that exactly models the time evolution of susceptible-infectious dynamics on finite-size networks. The equivalent process has only a small number of dynamical variables, offering tremendous computational advantages in terms of memory requirements. For the same reason, the stochastic equation leads to an analytical Gaussian approximation quite precise even for rather small networks and valid for all times. Furthermore, our approach generalizes to a vast class of systems.