Marcos Goycoolea, Associate Professor, School of Business, Universidad Adolfo Ibañez

Friday, May 21, 16h00, IMB 385

Gomory Mixed Integer (GMI) cuts are among the most effective cutting planes used for solving mixed integer programming (MIP) problems in software packages. Traditional GMI cut algorithms only generate these from the optimal bases of LP relaxations. Instead, we consider other (non-optimal and possibly infeasible) bases. This allows us to generate many consecutive rounds of rank-1 GMI cuts, which are both effective and sparse. Also, it allows us to apply GMI cuts in new contexts, such as branch-and-bound, and others.

This is joint work with Sanjeeb Dash.