Key points are not available for this paper at this time.
Abstract The procedure which involves successively minimizing a convex function with respect to each of its coordinates is shown to converge if constraints are rectangular and the function has continuous derivatives. It is also shown that certain more general procedures, which might be expected to converge more rapidly, are convergent. A criterion is exhibited which yields information concerning the rate of convergence of this method.
D. A. D'Esopo (Sun,) studied this question.