Abstract: Self organizing systems involve a collection of
dynamically interacting components such that the individual rules of
interaction among components result in an interesting collective
configuration. In controlled self organizing systems, the objective is
to *design* rules that provably lead towards desirable configurations.
An important concept in such systems is stochastic stability. In
contrast to deterministic convergence, stochastic stability
characterizes the most likely set of collective configurations under
perpetual random evolution. This talk presents an approach to
controlled self organization from the perspective of game theoretic
learning. In particular, the talk presents a collection of interaction
rules and accompanying stochastic stability analysis for i) distributed
adaptation and cooperative control under constrained evolution and ii)
self assembly under constrained communication.