A Comparison of the Primal-Simplex and Primal-Dual Algorithms for Linear Programming A statistical comparison of the primal-dual and the more commonly used primal-simplex algorithm for solving linear programming problems has been made under the assumption of starting with a full artificial basis. Under these conditions the primal-dual method shows a statistically significant superiority on randomly generated problems. It has also been found, via a regression analysis, that the relevant parameters in determining the difference in the number of iterations between the algorithms is not only the number of constraints and the number of variables but also the ratio of the latter to the former. CACM November, 1965 Mueller, R. K. Cooper, L. CA651110 JB March 6, 1978 4:50 PM 1171 5 1171 1171 5 1171 1171 5 1171