A Context-Aware Cutting Plane Selection Algorithm for Mixed-Integer Programming

It is used to tighten the Linear Programming (LP) relaxation of (1). Cuts are generated in rounds, where we refer to the process of computing cuts, applying a subset of the computed cuts, and resolving the LP as a separation round. In general, cuts are cheap to compute, and more are computed than actually applied to the LP. The process of deciding which cuts to add to the LP is called cut selection, and is the focus of this paper, in which we introduce a new cut selection algorithm. Cut selection was considered largely "solved", following computational results on a variety of MILP solvers [1, 2, 3], in that cheap heuristic selection rules were sufficient for good performance.