摘要
In this paper, we propose a finite barrier kernel function for primal-dual interior-point algorithm in linear optimization with a full-Newton step. To our best knowledge, it is the first time that the property of exponential convexity is used for full-Newton step interior-point methods(IPMs). Moreover, the analysis is simplified and the complexity of the algorithm coincides with the currently best iteration bound for linear optimization problems.
- 出版日期2016-11
- 单位中国人民解放军空军工程大学; 西安电子科技大学; 西安工业大学