一类光滑罚算法的全局收敛性

Global convergence of a class of smoothing penalty function

对约束优化问题给出了一类光滑罚函数,它是根据一类光滑逼近精确罚函数的光滑函数而提出来的,并基于这类罚函数给出了罚算法.在很弱的条件下,建立了光滑罚算法的一个摄动定理,由此定理得到该罚算法的全局收敛性.

A class of smooth penalty function is proposed for the constraint optimization problem,which is based on the smooth function of a class of smooth approximation precision penalty function,and the penalty algorithm is given based on this kind of penalty function.In the very weak condition,a perturbation theorem of the smooth penalty algorithm is established,and global convergence of the penalty algorithm is obtained.