一个新的简单精确光滑罚函数

A New Simple Exact and Smooth Penalty Function

针对一般约束优化问题,通过添加一个变量,给出一个新的简单精确光滑罚函数.在较弱的约束品性的条件下,证明所给出的罚函数具有一定的连续可微性,而且当罚参数充分大时,所给出的罚问题的局部极小点为原问题的局部极小点.

By adding one variable, a new simple exact and smooth penalty function is proposed for general constrained optimization problems. Under weaker constraint qualification assumptions, it is proved that when the penalty parameter is sufficiently large, the local minimizer of this penalty function is the local minimizer of a primal problem.