2783. Rotational relaxations for global optimization of quadratic programs
Invited abstract in session WB-35: MINLP 1, stream Continuous and mixed-integer nonlinear programming: theory and algorithms.
Wednesday, 10:30-12:00Room: Michael Sadler LG15
Authors (first author is the speaker)
| 1. | Yehor Blokhin
|
| Mathematics, KTH Royal Institute of Technology | |
| 2. | Jan Kronqvist
|
| Mathematics, KTH Royal Institute of Technology |
Abstract
In this talk, we consider the global optimization of non-convex quadratic programs with linear constraints, defining a bounded set. We introduce a family of new relaxation methods, which are derived by reformulating the problem by linearly transforming the vector space using constraints. The preliminary results show that the presented approaches, integrated with the αBB method and McCormick envelopes, can massively increase the quality of the convex relaxation.
Keywords
- Programming, Quadratic
- Continuous Optimization
Status: accepted
Back to the list of papers