ISE Seminar Series: MIP-based Algorithms for Global Optimization of MINLPs
This event is in the past.
This is part of the Department of Industrial & Systems Engineering Seminar Series.
This talk will introduce a fully MIP-based algorithm to solve MINLPs to global optimality, as an alternative to the traditional spatial branch-and-bound approach. This work has three major ideas which will be presented in detail. They are building disjunctive relaxations for multilinear terms, disjunctive relaxations for nonlinear univariate functions, and adaptive multi-variate partitioning algorithm. These ideas will be combined to develop the MIP-based algorithm that has proof of convergence to global optimality. Extensive computational experiments on benchmark instances will also be presented and pros and cons of this method will be discussed.
Kaarthik Sundar is a staff-scientist in the Information Systems and Modeling Group at Los Alamos National Laboratory. Prior to that, he was a post-doctoral researcher at the Center for Nonlinear Studies in Los Alamos National Laboratory in Los Alamos, New Mexico. He earned his Ph.D. and M.S. in Mechanical and Electrical Engineering, respectively, at Texas A&M University, College Station, Texas. His research interests broadly include mathematical programming and numerical optimal control applied to areas of transportation, surveillance, and interdependent energy infrastructure systems.