EECS 495: Distributed Optimization

Quarter Offered

Winter : 2-3:20 TuTh ; Wei


One course in optimization (including primal, dual problems, Lagrangian functions), linear algebra, calculus, familiar with real analysis or permission of instructor.


CATALOG DESCRIPTION: Large-scale networks and datasets, coming from applications such as Internet, wireless sensor networks, robotic networks and large scale machine learning problems, are an integral part of modern technology. One main characteristic of these systems is the lack of centralized access to information due to either communication overhead or the large scale of the network. Therefore control and optimization algorithms deployed in such networks should be completely distributed, relying only on local information and processing. This course studies various models and algorithms in the distributed and parallel settings. Topics include graph theory, algorithms for solving linear equations, iterative methods for convex problems, synchronous and asynchronous setups, consensus algorithms and rate analysis.

REQUIRED TEXT: Parallel and Distributed Computation by Dimitri Bertsekas J Tsitsiklis

GRADING: Grades will be based on bi-weekly homework, midterm and a course project