Workshop 6

Advances in Distributed and Large-Scale Optimization


Mikael Johansson, KTH Royal Institute of Technology
Angelia Nedich, Arizona State University


We are witnessing an extraordinary growth in a variety of networked systems, including sensory systems, power systems, communication networks, health systems, and data management systems, as well as hybrid systems arising from virtual and physical integration of such systems. A multitude of issues arise with control and optimal operations of such networked systems, including performance guarantees, resilience, and reliability. Distributed optimization methods play a crucial role in addressing a variety of problems arising in interconnected systems. Significant research efforts in the development of such computational methods have been dedicated recently in a multitude of directions ranging from new theoretical advances in support of the design of new optimization approaches (such as splitting techniques), the development of algorithms with improved efficacy and scalability, to dealing with privacy concerns and communication efficiency.

The goal of this workshop is to provide a collection of Talks that showcase some recent developments and applications of decentralized and/or distributed computational methods in optimization and control of emerging large scale and distributed networked systems.


Morning Session I:
– 09.00-09.45: Proximal envelopes, Andreas Themelis and Panos Patrinos, KU Leuven, Belgium
– 09.45-10.30: Sparsity and randomisation in large-scale convex optimization problems, Alexander Gasnikov, Moscow Institute of Physics and Technology, Moscow, Russia

10.30-11.00: Coffee break

Morning Session II:
– 11.00-11.45: Novel tools for analyzing asynchronous optimization algorithms,
Hamid Reza Feyzmahdavian and Mikael Johansson, ABB Corporate Research and KTH, Stockholm, Sweden
– 11.45-12.30: Convergence of asynchronous sub gradient push under bounded delays, Mahmoud Assran and Mike Rabbat, McGill University and Facebook Research, Motreal, Canada

12.30-14.00: Lunch break

Afternoon Session:
– 14.00-14.45: Coordinate descent methods and solving Laplacian systems, Mert Guruzbalaban, Rutgers University, USA
– 14.45-15.30: On optimal convergence rates of distributed algorithms, Angelia Nedich, University of Arizona, USA