1: \begin{abstract}

2: The success of compressed sensing relies essentially on the ability to efficiently find

3: an approximately sparse solution to an under-determined linear system. In this paper, we developed

4: an efficient algorithm for the sparsity promoting $\ell_1$-regularized least squares problem

5: by coupling the primal dual active set strategy with a continuation technique (on the regularization

6: parameter). In the active set strategy, we first determine the active set from

7: primal and dual variables, and then update the primal and dual variables by solving a

8: low-dimensional least square problem on the active set, which makes the algorithm very efficient.

9: The continuation technique globalizes the convergence of the algorithm, with provable global convergence

10: under restricted isometry property (RIP). Further, we adopt two alternative methods, i.e., a modified discrepancy

11: principle and a Bayesian information criterion, to choose the regularization parameter. Numerical

12: experiments indicate that our algorithm is very competitive with state-of-the-art algorithms in

13: terms of accuracy and efficiency.

14: \end{abstract}