Paper
Preprint on separating hyperplanes for polynomial MINLP
Separating hyperplanes from the moment and sums-of-squares hierarchy, added to the outer approximation for mixed-integer polynomial optimization problems.
A new preprint on Optimization Online uses the moment and sums-of-squares hierarchy to compute separating hyperplanes for nonconvex mixed-integer polynomial optimization problems, and embeds them in a polyhedral outer approximation framework. It is an example of the external cut generation that the new hyperplane callback in SHOT 2.0 makes possible.
Eggen, C., Kronqvist, J., Lundell, A. and Volkwein, S. Separating hyperplanes for mixed-integer polynomial optimization problems. Optimization Online (2026). optimization-online.org/?p=33712