1: \begin{abstract}

2: We present a new algorithm and the corresponding convergence analysis for the 

3: regularization of linear inverse problems with sparsity constraints, applied to a 

4: new generalized sparsity promoting functional. The algorithm 

5: is based on the idea of iteratively reweighted least squares, reducing the minimization 

6: at every iteration step to that of a functional including only $\ell_2$-norms. 

7: This amounts to smoothing of the absolute value function that appears in the 

8: generalized sparsity promoting penalty we consider, with the smoothing becoming 

9: iteratively less pronounced. We demonstrate that the sequence of iterates of our 

10: algorithm converges to a limit that minimizes the original functional. 

11: \end{abstract}

12: