A New Cutting Plane Algorithm for a Class of Reverse Convex 0-1 Interger Programs

A new algorithm is presented for a wide class of optimization problems where some of the objective function and constraints are nonlinear and where the variables are 0-1 integer. The algorithm is based on a cutting plane method. The cutting plane is constructed using the level sets of the functions involved as well as the edges of the polytope defining the linear constraints. We show that the algorithm converges finitely to a global optimum of the non-linear/0-1 integer optimization problem.