Pengembangan Algoritma Frank-Wolfe Untuk Menyelesaikan Problema Quadratic Integer Programming Murni

Abstract

Quadratic Integer Programming is a mathematical programming model in which the objective function has quadratic form, the constraints are linear and the decision variables are restricted to integer value. Such model has many applications, particularly in finance, network design, chemical processing and scheduling. This thesis proposes a method developed for Frank-Wolfe method for solving the quadratic integer programming problem. Frank-Wolfe method is an iterative method which is used particularly for solving quadratic programming problem. The search direction vector concept found in Frank-Wolfe method could be developed in such a way would direct an optimal continuous solution to feasibel integer point. Afterward an integer step size is used to compare the value of objective function started from the feasible integer point which has just obtained.