Abstract
Statistical physics of spin systems provides a framework to investigate mechanisms underneath information propagation on complex networks. This thesis strengthens the bond between the branch of physics and network science and improves the developing framework. Specifically, this thesis generalizes the topology of spin systems from regular lattices to various types of complex networks and studies the spin statistics in the systems with numerical simulation. After reviewing fundamental statistical physics and graph theory, subsequent chapters analyse the topological dependence of the systems’ thermal criticality and the extent of damage by freezing spins with various centrality—a spin system’s critical temperature is lower on a smallworld network but higher on a scale-free network, and it is more destructive to a spin system by freezing a spin with a high degree or a high information centrality. Lastly, the thesis extends centrality measures to measures of efficiency in information
propagation, like joint sales potential, and discusses an algorithm that may locate a group of nodes that has a high joint sales potential.