Abstract
This paper describes an interior-point algorithm for linear programming that is almost as simple as the affine-scaling method and yet achieves the currently best complexity of O(root n t) iterations to attain precision t. The basic algorithm needs neither dual estimates nor lower bounds, although its analysis is based on Ye's results for the primal-dual potential function. Some computationally preferable variants are also presented.