摘要
锥规化模型在二次规化和组合最优化中十分重要。带有线性约束和0-1约束的二次规化问题都可以改写为一个锥规化问题,但锥规化是一个NP难问题,本文考虑在标准单形上定义某种特殊的二阶锥,利用这种锥去近似逼近一种特定形式的锥规化的解,得出:标准单形划分越细,逼近这种锥规化解的误差也越小。
Cone programming is very important in quadratic program and combination optimization. The quadratic program with linear or 0-1 constrain can be simplified to cone programming. But,cone programming is a NP-hard problem. In this paper,a second-order cone is considered,which is defined on the standard simplex,and this cone is used to come close to the standard form cone programming. As the simplex partition gets finer and the error of the approximation becomes smaller.
作者
李扬
顾世煜
LI Yang;GU Shiyu(Shenyang Ligong University,Shenyang 110159,China;The middle school of Northeast Zhongshan,Shenyang 110001,China)
出处
《沈阳理工大学学报》
CAS
2018年第3期91-94,共4页
Journal of Shenyang Ligong University
基金
沈阳理工大学青年教师启动专项基金(QN201603)
关键词
锥规化
标准单形
对偶锥
半正定规化
cone programming
standard simplex
dual cone
semidefinite programming
