Faculty Contact: Jason Marden
In the design of distributed optimization algorithms, it is sought that individual agents, by trying to minimize their "local" objective function, can indeed minimize a desired "global" objective function. An important question is the mapping or correspondence of "optimality" between the global and the local objective functions, and it is a major theoretical work to guarantee that this correspondence may be preserved under different conditions over the agents, e.g., when agents have restricted communication or information exchange among them. One framework for designing such a mapping is through Game Theory; however, current literature ensures the existence of such mapping under (strong) assumptions of the structure of the global objective function (e.g., "smoothness", convexity, etc.). or have been proved for specific classes of functions. This module's objective is to be an introductory literature review of those current problems and identify further open problems in the field of network games and distributed optimization.
- Fall 2017: Pedro Cisneros