摘要
带自由变量的广义几何规划(FGGP)问题广泛出现在证券投资和工程设计等实际问题中.利用等价转换及对目标函数和约束函数的凸下界估计,提出一种求(FGGP)问题全局解的凸松弛方法.与已有方法相比,方法可处理符号项中含有更多变量的(FGGP)问题,且在最后形成的凸松弛问题中含有更少的变量和约束,从而在计算上更容易实现.最后数值实验表明文中方法是可行和有效的.
Generalized geometric programming (FGGP) frequently in portfolio investment and engineering design problems with free variables occur By utilizing equivalent transfor marion and the convex underestilnate of the objective and constraint functions, a convex relaxation method is proposed for finding global solution of (FGGP). In comparison with the method presented, this approach can solve signomial terms with more variables of (FGGP), and the convex relaxed problem produced involves less variables and constraints, so it can be realized more easy in computation. The numerical experiments show the feasibility and efficiency of the proposed method.
出处
《数学的实践与认识》
CSCD
北大核心
2012年第12期100-106,共7页
Mathematics in Practice and Theory
基金
河南省教育厅自然科学研究计划项目(2011B110012
12B110004)
关键词
广义几何规划
自由变量
全局解
凸松弛
generalized geometric programming
free variables
global solution
convex re-laxation