Margarita Paz Castro, University of Toronto. “Relaxed Decision Diagrams for Discrete Optimization Problems”

Position:  PhD Candidate

Current Institution:  University of Toronto

Abstract:  Relaxed Decision Diagrams for Discrete Optimization Problems

Discrete optimization is one of the main research fields in operational research and computer science due to its challenging problems and many applications. Several techniques have been used to solve these problems, such as integer programming, constraints programming, and heuristic search. Relaxed Decision Diagrams (DDs) are one of the newest approaches to tackle discrete optimization problems, which use a graphical structure to represent the set of feasible solutions and compute bounds. My research aims to investigate and extend the use of relaxed DDs as a tool to solve discrete optimization problems across different fields. I want to uncover the advantages of the approach, its relationship to existing techniques, and its uses in conjunction with other approaches, such as integer programming. I am currently working with challenging (NP-hard) problems from the operations research and artificial intelligence (AI) communities, including vehicle routing and scheduling and AI planning problems.

Margarita Castro is a fourth-year PhD candidate in the Mechanical and Industrial Engineering Department at University of Toronto. She is part of the Toronto Intelligent Decision Engineering Laboratory (TIDEL) and is under the supervision of Professors J. Christopher Beck and Andre Cire. She received a full PhD scholarship from the Chilean government. She received a master’s degree in 2014 from the Pontificia Universidad Catolica de Chile, where she graduated with maximum distinction. Her research focuses on discrete optimization problems and the use of relaxed decision diagrams to solve them. She is interested in constraint programming, integer programming, stochastic programming, AI applications, machine learning, scheduling, and vehicle routing problems.