TechNews Pictorial PriceGrabber Video Mon Apr 23 05:52:32 2018


Strong polynomiality of the simplex method for totally unimodular linear program
Source: Tokyo Institute of Technology

The simplex method is a practical and efficient algorithm for solving linear programming problems, but it is theoretically unknown whether it is a polynomial or strongly-polynomial algorithm.

É. Tardos of Cornell University proposed a strongly polynomial algorithm for combinatorial linear programming problems and T. Kitahara and S. Mizuno at Tokyo Institute of Technology have obtained a new bound for the number of distinct solutions generated by the simplex method.

It is known that the simplex method requires an exponential number of iterations for some special linear programming instances. Hence the method is neither polynomial nor a strongly-polynomial algorithm for general linear programming problems.

The researchers showed that the simplex method using Tardos' basic algorithm is strongly polynomial for totally unimodular linear programming problems, if the problems are nondegenerate.

These findings give a theoretical background as to why the simplex method could solve a large scale linear programming problem efficiently.

© 2018 PopYard - Technology for Today!| about us | privacy policy |