1: \begin{abstract}

2: We study the convergence of the gradient descent method for solving 

3:  ill-posed problems where the solution is characterized

4: as  a global minimum of a differentiable functional in a Hilbert space. 

5: The classical least-squares functional for nonlinear operator equations is 

6: a special instance of this framework and the gradient method then reduces to 

7: Landweber iteration. The main result of this article is a  

8: proof of weak and strong convergence under new nonlinearity conditions that

9: generalize the classical tangential cone conditions. 

10: \end{abstract}

11: