SOLUTION OF GEOMETRIC PROGRAMS VIA SEPARABLE PROGRAMMING

Problem of choosing the best decision from a number of feasible alternatives, the optimal one being that particular decision which is most attractive based on some measure of effectiveness - such as profits, costs, time or volume. Recently, a relatively general branch of non-linear programming, called geometric programming, has been developed. Application of geometric programming theory spans many of the classical subsets of non-linear programming. However, to date, large geometric programs have been difficult to solve. This paper presents a simplex-based approach to the solution of such programs which is indifferent to the degree of difficulty of the problem. Furthermore, unlike many gradient-search techniques its success is not dependent upon a good starting point.

Overview