Infeasible-Interior-Point Algorithms

摘要: An interior-point algorithm whose initial point is not restricted to a feasible called an infeasible-interior-point algorithm. The directly solves given linear programming problem without using any artificial problem. So the has big advantage of implementation over feasible-interior-point algorithm, which start from point. We introduce primal-dual and prove global convergence When all data are integers, terminates in polynomial-time under some moderate conditions also predictor-corrector achieves better complexity superlinearly convergence.