Seminar: MWU 2.0 with approximation guarantee for non-convex (structured) (MI)NLPs

Luca Mencarelli, a researcher in optimization, has recently joined the department and our laboratory. To welcome him, we have invited him to give a talk on his research. Everyone is welcome to join.

Date/Time: Wednesday, May 3 Monday, May 8, 2023, 16:00. Date changed!

Room: Sala Seminari Ovest, Dipartimento di Informatica.

Title: MWU 2.0 with approximation guarantee for non-convex (structured) (MI)NLPs.

Speaker: Luca Mencarelli, Department of Computer Science, University of Pisa.

Abstract: In this talk, we introduce a new approximation guaranteed Multiplicative Weights Updated (MWU) for Mixed Integer Nonlinear Programming (MINLP). We introduce the general MWU framework and we describe its application to two hard global optimization problems, namely the Distance Geometry Problem (DGP) and the (non-convex) Separable Programming (SP). Preliminary yet promising computational results show the potential of the proposed algorithm in terms of efficacy and efficiency for the previous problems.