PhD Thesis Final Defense to be held on September 17, 2019, at 9:00
Photo credit: Stratis Skoulakis
The examination is open to anyone who wishes to attend (Multimedia amphitheater, Central Library of NTUA)
Thesis Title: Natural and Efficient Dynamics through Convex Optimization
Abstract: This thesis studies one-shot games, discrete-time dynamical systems and combinatorial optimization problems relating to the way people form their opinions. We use ideas and techniques developed in the area convex optimization to examine the dynamic behavior of the opinions when selfish agents iteratively play such opinion formation games and to bound the inefficiency of equilibrium in terms of total disagreement cost. All the considered opinion dynamics are natural in the sense that they are based on widely accepted game playing strategies such as best response or no-regret algorithms, while at the same time they are efficient in the sense that they converge fast to well-behaved equilibria even with limited information exchange among the agents. Last but not least, we use convex optimization techniques to study a dynamic version of the k-median problem in which the requests change their positions over time. This problems relates to the way a political party should assign public positions to its representatives so as to efficiently cover voters with dynamically changing opinions over time.
Supervisor: Dimitris Fotakis, Assoc. Professor
PhD student: Stratis Skoulakis