Research

Faster than simulated annealers and rivaling the performance of quantum annealers, the Fujitsu Co-Creation Research Laboratory applies quantum-inspired Digital Annealer technology to solving combinatorial optimization problems.

Professor Christopher Beck

Our research on the DA focuses on understanding how it can be used to solve hard combinatorial problems in Artificial Intelligence and Operations Research. We are interested in a variety of problem classes (e.g., constraint programs, SAT) and applications such as scheduling and wind farm layout optimization. We will seek to use the DA to find exact solutions through hybridization with existing exact optimization methods and to expand the size of problems that can be addressed through systematic decomposition frameworks.

Professor Mojgan Hodaie

Gamma Knife Radiosurgery (GKRS) is a non-invasive treatment modality for the focal irradiation of brain tumours and nerve abnormalities. It involves a stereotactic frame and helmet array of 192 gamma sources, which can be separately controlled to precisely contour the prescribed radiation dose to the target while minimizing exposure to surrounding healthy tissue. Each GKRS plan is a unique combination of many input variables and is thus well-formulated as an optimization problem for the DA. The DA will be used to generate precise and personalized GKRS treatment plans in short timeframes to enhance the patient journey.

Professor Jeffrey Rosenthal

We study the theoretical and mathematical properties of Markov chain Monte Carlo (MCMC) and annealing algorithms.  We are investigating the ways in which the DA uses these algorithms, and how they can be improved to produce more efficient optimizations and more accurate samples. We have created a Javascript web page to illustrate some of these issues, which can be viewed at: http://probability.ca/metropolis

Professor Edward Sargent

Our research focuses on the application of DA for material design specifically to optimize components and ratios of high-entropy alloys for water splitting using DA. We are developing a machine learning method with uncertainty prediction in the form of quadratic unconstrained binary optimization (QUBO). Based on this machine learning method, we are designing high-entropy alloys for water-splitting catalytic materials using active learning and realizing the designed catalysts experimentally.

Professor Daryl Nazareth

One of the most prevalent cancer treatments is called Volumetric Modulated Arc Therapy (VMAT). This modality allows the tumour to be irradiated with high-energy radiation beams, generated by a medical linear accelerator, while the dose delivered to the surrounding healthy tissue is minimized. VMAT treatment planning is therefore an optimization problem involving thousands (or even tens of thousands) of independent variables, which must be solved within reasonable clinical timeframes. This makes the DA ideally suited for the VMAT planning problem, and involves an interdisciplinary collaboration between medical physics, engineering, and computer science.