Reformulating mixed-integer quadratically-constrained quadratic programmes
Prof. Adam Letchford
Lancaster University, UK
DEI - Conference Room
March 30th, 2011
It has been known for some time that semidefinite programming (SDP) can be used to form useful relaxations of many different NP-hard optimisation problems. In a series of recent papers, Billionnet and colleagues have proposed to use SDP to reformulate mixed-integer quadratic programs, rather than merely relax them. We will review this work, and then examine conditions under which the idea can be extended to problems with quadratic constraints. The talk is based on joint work with Laura Galli from the University of Bologna.
Applied systems analysis and operations research