A combined algorithm for fractional programming (Q5959292)

The problem considered is that of mazimizing the quotient of two d.c. (difference of convex) functions over a convex, compact subset of \(\mathbb{R}^{n}\); the numerator and the denominator are assumed to be nonegative and strictly positive, respectively. The solution method proposed starts with transforming the problem into another constrained optimization problem, in which both the objective functions and the constraints are d.c.. The new problem is then solved by a branch-and-bound algorithm, which generates cutting planes to approximate the feasible region as well as conical partitions. The algorithm is shown to converge, and its behavior is illustrated by a numerical example, from the results of which the author draws the conclusion that the algorithm is likely to solve small problems efficiently. He also discusses about the possibility of extending the method for solving the sum-of-ratios problem.