1: \begin{abstract}

2: 	We present a method 

3: 	to solve a special class 

4: 	of parameter identification problems

5: 	for an elliptic optimal control problem

6: 	to global optimality. 

7: 	The bilevel problem is reformulated via

8: 	the optimal-value function of the lower-level problem.

9: 	The reformulated problem 

10: 	is nonconvex and standard regularity conditions 

11: 	like Robinson's CQ are violated. 

12: 	Via a relaxation of the constraints,

13: 	the problem can be decomposed

14: 	into a family of convex problems

15: 	and this is the basis for

16: 	a solution algorithm.

17: 	The convergence properties are analyzed.

18: 	It is shown that a penalty method can

19: 	be employed to solve this family of problems

20: 	while maintaining convergence speed.

21: 	For an example problem, the use of the

22: 	identity as penalty function allows

23: 	for the solution by a semismooth Newton method.

24: 	Numerical results are presented.

25: 	Difficulties and limitations 

26: 	of our approach to solve a nonconvex problem to

27: 	global optimality are discussed.

28: \end{abstract}

29: