## A nonmonotonic trust region technique for nonlinear constrained optimization.(English)Zbl 0822.65039

This paper describes a nonmonotone trust region method for equality constrained optimization problems. A nonsmooth penalty function is introduced as the merit function, and a correction term is added to the step in order to overcome the Maratos effect. It is shown that all accumulation points of the resulting sequence are Kuhn-Tucker points and its local convergence rate is $$q$$-superlinear under certain assumptions.
Reviewer: P.Pan (Nanjing)

### MSC:

 65K05 Numerical mathematical programming methods 90C30 Nonlinear programming