On the solution of a two ball trust region subproblem

The paper investigates the structure of a two ball trust region subproblem arising frequently in nonlinear parameter identification problems and propose a method for its solution. The method decomposes the subproblem and allows the application of efficient, well studied methods for the solution of trust region subproblems arising in unconstrained optimization. In the discussion of the structure the paper focuses on the case where both constraints are acive and on the treatment of the unconstrained problem.