The myriad things rise and fall, I thereby observe and contemplate the Law.All things bloom and flourish, then return to their roots, each by each.Laozi "Dao De Jing" (400 BC) knowledge on the behavior of a single starling does not ensure the comprehension of their flocking behavior. Fascinating phenomena such as the weather, life, and the flocking of birds arise from large collections of simpler components, but are much more than the sum of them: the dynamical interaction between the parts gives rise to the complexity of the whole. media [34]. Novel classes of patterns emerge if the underlying network architecture is further generalized. Patterns such as travelling waves and quasi-stationary patterns can be induced in directed, hence non-symmetric networks [35] and distinct heterogeneous patterns occur in multiplex networks [36].Thus, we can discern that, despite the importance and ubiquity of the problem of perturbation spreading, general answers to simple questions like "When and how will a perturbation affect a given node?" are still missing to date. Even for networks of linear deterministic dynamical systems, the questions remain open [64].
Synopsis of the thesisIn this thesis, we develop a theory of dynamic response patterns for complex networks under the influence of fluctuating perturbations. Motivated by the functional robustness
Part I. FundamentalsChapter 1
Networks as Dynamical SystemsAs the beginning of the thesis, we present the most fundamental concepts underlying the study of dynamic network responses in this thesis. We start with basic and relevant concepts in graph theory, such as graphs, paths and distances (Sec. 1.1). Particularly, we highlight the relation between the distance between a pair of nodes and the element of the power of Laplacian matrices associated to the node pair. In Sec. 1.2, we move on to another branch of mathematics, i.e. dynamical system theory. We present the definition of dynamical systems and focus on the linear stability analysis at fixed points. In the last section of the chapter, Sec. 1.3, we combine the aforementioned concepts from the two branches and present basic considerations about the dynamics of a networked system, i.e. a system of coupled dynamical systems. We especially focus on the linearization of the dynamics of a general network system with pair-wise coupling at the fixed point.
NetworksNetworks are closely related to the concept graph in mathematical literature, which is defined as a collection of vertices connected by edges, representing how a set of objects are related to each other (Fig. 1.1). Vertices and edges are sometimes also called nodes and links, or sites and bonds [4]. The three terms "vertices", "nodes" and "sites" referring to the same notion are interchangeable in this thesis, as well as "edges", "links", and "bonds".
The matrix representation of graphsMathematically, a graph G(V, E) consists of a non-empty finite set V of vertices and a finite family E of edges, which are unordered pairs of vertices [5]. In this thesis we consider only ...