Claudia D’Ambrosio, Dipartimento di Elettronica, Informatica e Sistemistica, Università di Bologna

Tuesday, January 18, 11h00, IMB 385

Starting from a real-world application, the Unit Commitment and Scheduling of Hydroelectric Power Plants, we review different standard methods that have been traditionally used to approximate nonlinear functions of two variables. In particular, piecewise linear approximation and simplicial decomposition, and different possible ways to model them as mixed integer linear programming problems. Moreover, two new approaches are presented, the first being an improvement of the standard piecewise linear approximation, and the second called an optimistic mixed integer linear programming modeling method. Advantages and drawbacks are presented for each method, also within the context of the real-world application.

The talk is based on joint work with Alberto Borghetti, Andrea Lodi, Silvano Martello and Riccardo Rovatti.