提出了一种新的内点算法——宽邻域预估校正算法。该算法基于经典预估校正算法思想,把窄邻域拓展到宽邻域里,使算法更快地迭代。给出了算法的具体步骤,讨论了其计算复杂性,分析结果表明,所给算法是一多项式时间算法。通过数值实验验证算法的有效性。
A new interior point wide-neighborhood predictor-correcting algorithm is presented for a linear programming problem. On the basis of the idea of a predictor-correcting algorithm, the iteration of our algorithm is faster in a wide-neighborhood than in a narrow one. The concrete steps of the algorithm are introduced, its computational complexity is discussed, and the results indicate that the algorithm is a polynomial-time one. The validity of the algorithm is confirmed though a numerical experiment.