Computational Methods in Industrial Mathematics

Ryerson University

Current economic development is based on continuous increase in efficiency of business processes and industrial organization. This increase is obtained through better collection of business process data, data analytics tools (including machine learning and data science methods), and advanced simulation and optimization techniques. Including available data within optimization of processes requires constructing complex mathematical models. In particular, many real world scenarios exhibit high combinatorial complexity, need to include several sophisticated constraints imposed by the business environment, and need to consider uncertainty. Including all key features characterizing industrial processes and their business environment leads to complex mathematical models.

The complexity of the problems leads to a need for large, sophisticated computational systems, consisting of hundreds or thousands of computational units. Computational technologies (such as distributed computing, GPU computing, massive parallel computing) allow for the management of that process. Our lab tries to narrow the gap between the huge demand from the industrial side and the lack of tools on the academic side. Our goal is to focus on developing new mathematical methods and tools (new theorems, algorithms, and computer implementations of them, for non-standard industrial applications).