Probabilistic behavioral distance and tuning—reducing and aggregating complex systems

Abstract

Given two dynamical systems, we quantify how similar they are with respect to their interaction with the outside world. We focus on the case where simpler systems act as a specification for a more complex one. Combining a behavioral and probabilistic perspective we define several useful notions of the distance of a system to a specification.

We show that these distances can be used to tune a complex system. We demonstrate that our approach can successfully make non-linear networked systems behave like much smaller networks, allowing us to aggregate large sub-networks into one or two effective nodes. Finally, we discuss similarities and differences between our approach and H∞ model reduction.