Interior point methods specialized to the L∞ fitting problem are surveyed, improved, and compared with the traditional simplex approach. A primal affine-scaling interior point method is presented, completing the affine-scaling interior point family approach to the L∞ fitting problem. Computational complexity and data storage are reduced for interior point approaches when dealing with polynomial fitting problems. Numerical experiments indicate that interior point approaches rarely perform better than the simplex method for tested problems. The primal affine-scaling method presented in this paper achieved the best results among the interior point family.
