An extension of the Charnes-Cooper method in linear fractional programming

The method of Charnes and Cooper converts a linear fractional programming problem into a linear programming problem. This method assumes that the denominator of the objective function is positive on the feasible region and that the latter is bounded. This paper shows that even when these assumptions are discarded the method, if approrpiately interpretated, still works. This extends previous results of Mond and of Gavurin. An application to the computation of probability bounds is given.