1: \begin{abstract}

2: We consider descent methods for solving non-finite valued 

3: nonsmooth convex-composite optimization problems 

4: that employ Gauss-Newton subproblems to determine the iteration update. 

5: Specifically, we establish the global convergence properties for descent methods that

6: use a backtracking line search, a weak Wolfe line search, or a trust-region update.

7: All of these approaches are designed to 

8: %The descent algorithms are based on either a weak Wolfe or a backtracking line search using a continuous approximation to the directional derivative that 

9: exploit the structure associated with convex-composite problems.

10: \end{abstract}

11: