Smoothing augmented Lagrangian method for nonsmooth constrained optimization problems (Q496609)

A smoothing augmented Lagrangian algorithm is proposed to locate a stationary point of a general nonsmooth and nonconvex optimization problem. The convergence of the algorithm is established. The smoothing augmented Lagrangian method is applied to the bilevel program. The authors verify that either the exact penalty sequence is bounded or the weak generalized Mangasarian Fromovitz constraint qualification holds for all bilevel programs. Some numerical experiments for some general nonsmooth and nonconvex constrained optimization problems and some bilevel programs are reported.