Distributed Algorithms for Separable Optimization

In this talk we are interested in distributed algorithms for solving separable optimization problems. Many problems in engineering can be formulated as separable optimization problems, i.e., minimizing the sum of P functions subject to the intersection of P sets. Our goal is to solve such problem when the P functions and sets are not known at a single location, but rather distributed across a network with P nodes, each node having access to just one function and set. We present an algorithm based on the alternating direction method of multipliers that requires less communications than the state-of-the-art algorithms. Applications of this work include average consensus, distributed compressed sensing and SVMs, Internet protocols, and distributed model predictive control.