摘要
利用Fischer-Burmeister函数将混合互补问题转化为非线性方程组,由光滑函数逼近FB函数来求解非线性方程组.文中将信赖域方法和梯度法相结合,提出了Jacobian光滑化方法.算法在一定条件下的全局收敛性得到了证明,数值试验表明算法切实有效,有一定的优越性.
We converted the mixed complementarity problem into a system of nonsmooth nonlinear equations by using Fischer--Burmeister function, and we used a smooth function to approximate the Fischer- Burmeister function. By combining trust region method with gradient method, a Jacobian smoothing method was proposed. Under some conditions, we proved the global convergence and local convergence of the algorithm. Numerical result indicates that the algorithm is quite promising.
出处
《经济数学》
北大核心
2010年第3期73-78,共6页
Journal of Quantitative Economics
基金
湖南省教育厅资助项目(08C668)
关键词
混合互补问题
Jacobian光滑算法
信赖域方法
梯度步
全局收敛
二阶收敛
mixed complementarity problem
Jacobian smoothing method
trust region method
gradient step
global convergence
quadratic convergence
