Title: QP/NLP-based Branch-and-Bound algorithm for MINLP: It could work!

Abstract: In this talk, we discuss possible insights in order to deal with a QP/NLP-based Branch-and-Bound algorithm for Mixed Integer NonLinear Problems (MINLPs). After shortly presenting the Outer Approximation framework and the LP/NLP-based Branch-and-Bound algorithm for convex MINLPs, we introduce a convex quadratic under-approximation for convex functions, and, on the top of that, we draw a possible quadratic Outer Approximation scheme. Finally, we present preliminary computational results for convex and binary non convex MINLPs, by comparing the novel algorithm against state-of-the-art solvers.