Constraint Programming for Strictly Convex Integer Quadratically-Constrained Problems
2016
Inspired by the geometric reasoning exploited in discrete ellipsoid-based search (DEBS) from the communications literature, we develop a constraint programming (CP) approach to solve problems with strictly convex quadratic constraints. Such constraints appear in numerous applications such as modelling the ground-to-satellite distance in global positioning systems and evaluating the efficiency of a schedule with respect to quadratic objective functions. We strengthen the key aspects of the DEBS approach and implement them as combination of a global constraint and variable/value ordering heuristics in IBM ILOG CP Optimizer. Experiments on a variety of benchmark instances show significant improvement compared to the default settings and state-of-the-art performance compared to competing technologies of mixed integer programming, semi-definite programming, and mixed integer nonlinear programming.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
33
References
1
Citations
NaN
KQI