1: \begin{abstract}

2: 	In this paper, we revisit the classical problem of solving over-determined

3: 	systems of nonsmooth equations numerically. 

4: 	We suggest a nonsmooth Levenberg--Marquardt method for its solution which,

5: 	in contrast to the existing literature, does not require local Lipschitzness

6: 	of the data functions. This is possible when using Newton-differentiability

7: 	instead of semismoothness as the underlying tool of generalized differentiation.

8: 	Conditions for fast local convergence of the method are given.

9: 	Afterwards, in the context of over-determined mixed nonlinear complementarity systems,

10: 	our findings are applied, and globalized solution methods, based on a residual

11: 	induced by the maximum and the Fischer--Burmeister function, respectively, are constructed.

12: 	The assumptions for fast local convergence are worked out and compared.

13: 	Finally, these methods are applied for the numerical solution of bilevel optimization

14: 	problems. We recall the derivation of a stationarity condition taking the shape of

15: 	an over-determined mixed nonlinear complementarity system involving a penalty parameter, 

16: 	formulate assumptions for local fast convergence of our solution methods explicitly, 

17: 	and present results of numerical experiments. 

18: 	Particularly, we investigate whether the treatment of the appearing penalty parameter

19: 	as an additional variable is beneficial or not.

20: \end{abstract}

21: