Quadratic Optimization (QO) has been studied extensively in the literature due to its application in real-life problems. This thesis deals with two complicated aspects of QO problems, namely nonconvexity and uncertainty. A nonconvex QO problem is intractable in general. The first part of this thesis presents methods to approximate a nonconvex QP problem. Another important aspect of a QO problem is taking into account uncertainties in the parameters since they are mostly approximated/estimated from data. The second part of the thesis contains analyses of two methods that deal with uncertainties in a convex QO problem, namely Static and Adjustable Robust Optimization problems. To test the methods proposed in this thesis, the following three real-life applications have been considered: pooling problem, portfolio problem, and norm approximation problem.
|Qualification||Doctor of Philosophy|
|Award date||11 Dec 2017|
|Place of Publication||Tilburg|
|Publication status||Published - 2017|