Fast approximate solutions to large and sparse systems of equations via convex optimization

Many effective Convex Optimization techniques face a common bottleneck: the resolution of large, sparse systems of equations. Despite the important advances in the state-of-art at theoretical and algorithm development, a significant challenge persists in the tendency to overlook the commercial viability of these algorithms.
Recent strides in computer hardware, driven by advancements in AI and cryptocurrency, have made small clusters of GPUs or FPGAs viable for addressing high-value optimization problems encountered by Kinaxis customers. If these hardware configurations can significantly reduce computational time, they offer a promising solution.
The collaboration between uOttawa, Kinaxis, and MITACS aims to bridge the gap between academic research and industry in high-performance computing. This partnership focuses on developing efficient, sustainable, and scalable optimization methods and computational algorithms tailored to real-world challenges convex optimization.

Faculty Supervisor:

Augusto Gerolin;Aaron Smith

Student:

Partner:

Kinaxis Inc.

Discipline:

Mathematics

Sector:

Information and cultural industries

University:

University of Ottawa

Program: