A unifying algorithmic framework for structured optimization

From 01-01-2018 to 31-12-2021

Description

The explosive growth of computational power alongside the immense proliferation of devices have enabled solving once unforeseeable problems. At the same time, the Internet of Things is extraordinarily complex, involving huge amounts of data and interconnections of billions of smart devices. In this landscape, the thirst for novel theories is more than ever topical: to manage large-scale cyberphysical systems such as smart cities, it is integral to attack the problems cast by coordinating many computers via harvesting their enormous computational power. Optimization is the key to materialize this vision, being ubiquitous in all areas of engineering: embedded control, signal processing, data mining, machine learning, wireless networks and transportation. The deluge of modern optimization will lie in the ability to reliably and speedily solve huge-scale problems by distributing the computational burden to devices in charge of solving small problems in nanoseconds, using minimal power consumption and cross-device communication.
The aim of this project is to contribute in solidifying structured optimization as the linear algebra of the 21st century by developing a unifying algorithmic framework for a wide class of problems. We seek to design and implement algorithms with superior convergence rates and accuracy leveraging ongoing theoretical advancements by our research team for expanding traditional methods to address some of the biggest challenges of the information era.

Team

• Panagiotis Patrinos, Promoter

Financing

Funding: FWO - Research Foundation - Flanders

Program/Grant Type: FWO Project - FWO Research Project