一种新的低阶罚函数的光滑化方法

A New Smoothing Method for Lower-order Penalty Function

对于约束优化问题,给出了一种用二次连续可微函数光滑低阶罚函数的方法;在一些弱的假设条件下,证明了光滑后的罚优化问题的最优解是原优化问题的ε-近似最优解.

With regard to constrained optimization problems, this paper gives a kind of new method by using second-order continuous differentiable function to smooth lower order penalty function, under some weak supposed conditions, proves that the optimal solution of penalty and optimization problems after smoothing is 6-approximate optimal solution of original optimization problem.