IE-OPIM Joint Seminar 3 December 2014 at 13.40 at FENS G035
Completely Positive Optimization: Theory and Tractable Approximations
Emre Alper Yıldırım
Abstract: Completely positive optimization deals with the optimization of a linear functional over an affine subspace of the cone of completely positive matrices. Recently, it has been shown that every quadratic optimization problem with a mix of binary and continuous variables can be formulated as an instance of a completely positive optimization problem. Therefore, despite the convex nature of this class of optimization problems, the cone of completely positive matrices is computationally intractable. We discuss various tractable approximations of completely positive optimization problems. We present our results on the quality of polyhedral approximations on certain classes of quadratic optimization problems.
Bio: Emre Alper Yıldırım is an Associate Professor of Industrial Engineering at Koç University. He earned his B.S. degree in Industrial Engineering at Bilkent University in 1997 and his M.S. and Ph.D. degrees in Operations Research at Cornell University in 2000 and 2001, respectively. Prior to joining Koç University, he worked as a faculty member at Stony Brook University (SUNY) and at Bilkent University. His research interests are in the theory and applications of optimization and algorithm design and analysis. He has received the INFORMS Optimization Prize for Young Researchers in 2006, TUBITAK Incentive Award for Young Researchers in 2009, and the TUBA Award for Young Researchers in 2011. He currently serves as an Associate Editor for the journals Optimization Letters and Optimization Methods and Software..