By David Schlenker
The National Science Foundation (NSF) recently awarded University of Florida engineering professor Aleksandr Kazachkov $550,213 to develop integer optimization technology that may enhance power system operations, vehicle routing, and even organ transplant allocation.
The NSF Faculty Early Career Development (CAREER) program grant will fund research to improve the scalability, speed, and reliability of integer programming solvers via a new generation of learning-enabled cutting planes, which are constraints that can be automatically derived from an optimization instance to refine the search space of solutions.
“It began with a mathematical question: When and how do optimization solvers benefit from cutting planes that drastically differ from those currently employed by solvers?” said Kazachkov, an assistant professor with Industrial & Systems Engineering (ISE) and an assistant director of the Center for Applied Optimization (CAO).
The improvements, Kazachkov noted in his project summary, can enable solving challenging operational problems more quickly, with higher accuracy, or involving greater complexity. Through computational math and algorithms – tested on UF’s supercomputer, HiPerGator – the project aims to design algorithms to extend mathematical optimization solvers and facilitate efficiencies in real-world settings.
For example, given a set of patients in need of a kidney, each with a willing but incompatible donor, Kazachkov’s research may optimize the resulting “kidney exchange market” to improve patient outcomes. Or, in the energy sector, a system operator needs to choose which generators to use to meet estimated demand.
These decisions are made frequently, leading to similar optimization instances each time. This shared structure is the key opportunity for learning techniques in Kazachkov’s proposed methodology.
The project is tightly integrated with educational objectives to increase access of technical expertise for nonprofit and public sector organizations, who do not typically have in-house optimization staff, while fostering students’ long-term engagement in local communities through their analytics and engineering skills.
One such collaboration is to optimize pickup-and-delivery routes for a local food bank. This has already led to student involvement via senior design classes and undergraduate research opportunities.
Kazachkov was born in Russia and raised in Brooklyn, where his mother encouraged him to pursue medicine, while his father nudged him into a more hands-on engineering discipline. Navigating these influences, Kazachkov found a happy medium in operations research for his undergraduate studies at Cornell University.
“It is computational, mathematical, and applied,” he said. “Operations research is the science of designing better optimization methods and learning to improve decision-making across various domains.”