Solving MINLP polynomial problems to global optimality: RAPOSa solver

Abstract:
This talk introduces RAPOSa, a global solver for polynomial optimization, and presents its latest developments and ongoing research. The main focus is on its two most recent functionalities: i) the extension to solve mixed-integer MINLP problems and ii) the extension to solve problems on unbounded domains (in progress). With the inclusion of these two extensions, RAPOSa will be able to tackle any MINLP polynomial optimization problem and, eventually, solve it to certified global optimality.

The talk will delve into various aspects of some challenges faced in these extensions, including some unexpected findings regarding the effects of seemingly innocuous modeling choices.