 |
Center for Nonlinear Studies |
Power grids. Food webs. The Internet. All of these represent systems whose collective performance is determined by the networked interactions of their individual units. A fundamental property of networks is that perturbations to a small number of nodes can propagate to other nodes, possibly causing the system as a whole to change behavior or fail. In this talk I will discuss how the same principle can actually be exploited to prevent such cascading failures and, more generally, control network behavior. For example, a faulty or suboptimal metabolic network can often be healed by the deliberate suppression of specific enzyme-coding genes. Similarly, ecosystems damaged by invasive species might be restored to their pristine state by the judicious manipulation of certain indigenous populations. These compensatory perturbations can bring a system to a desired target state even when that state is not directly reachable, but identifying the exact modifications to the network that will produce the desired effect is a highly nontrivial task. I will show how the possibility of compensatory perturbations arises naturally from the nonlinear dynamics inherent to most real systems, and will present a systematic computational approach for their identification in a general network of dynamical units. I thus hope to convey how nonlinearity?commonly thought to be an obstacle to controlling natural and engineered systems?can actually be a blessing in disguise
Host: Aric Hagberg