Nonlinear Integer Programming and Discrete Optimization

Branch and bound has been widely recognized as an effective method in solving linear integer programming problems. This paper presents a study on the feasibility of the branch and bound method in solving general nonlinear mixed integer programming and discrete optimization problems. First, a description of the branch and bound method as applied to the integer case is given. Next, a computer code BBNLMIP is developed to carry out an experimental study on 22 test problems. The numerical results indicate the effect of the problem parameters such as number of integer variables and constraints. Finally, a method for extending the branch and bound principle to solve nonlinear discrete optimization problems is described.