Heuristic approach to the efficient control of complex networks

Large complex dynamical graphs behave in ways that reflect their structure. There are many applications where we need to control these graphs by bringing them from their current state to a desired state. Affecting the state of these graphs is done by communications with some key elements, called driver nodes. One of the structural characteristics of graphs is the number of driver nodes required to control it. Finding such a minimal set to control the whole network is a computationally prohibitive process. In many applications, quickly locating a small set of nodes that control a significant percentage of the network is a much better alternative. This is the approach we take in this thesis. We seek an efficient heuristic approach for identifying a small enough set of drivers that is effective in controlling most of the network. The heuristics are developed based on statistical characterization of driver nodes that are distinct from those of the rest of the network. We test the heuristic approach and discuss its effectiveness for a variety of networks.