1: \begin{abstract}

2: The iteratively reweighted $\ell_1$ algorithm is a widely used method for solving various regularization problems, 

3: which generally minimize a differentiable loss function combined with a convex/nonconvex 

4: regularizer to induce sparsity in the solution. 

5: However, the convergence and the complexity of iteratively reweighted $\ell_1$ algorithms is generally difficult 

6: to analyze, especially for non-Lipschitz differentiable regularizers such as $\ell_p$ norm regularization with 

7: $0<p<1$.  In this paper,   we propose, analyze and test a reweighted $\ell_1$ algorithm combined with 

8: the extrapolation technique under the assumption of \emph{Kurdyka-\L ojasiewicz} (KL)  property  on the objective. 

9: Unlike existing iteratively reweighted $\ell_1$ algorithms with extrapolation, our method does not require 

10: the Lipschitz differentiability on the regularizers nor   the smoothing parameters in the weights 

11: bounded away from 0.    We show the proposed algorithm converges uniquely 

12: to a stationary point of the regularization problem and has local linear complexity---a much stronger result than existing 

13: ones.  Our numerical experiments show the efficiency of our proposed method.

14:  

15: \noindent{\bf Keywords:}  {$\ell_p$ regularization \and extrapolation techniques \and iteratively reweighted methods \and \emph{Kurdyka-\L ojasiewicz} 

16: \and non-Lipschitz regularization}

17: % \PACS{PACS code1 \and PACS code2 \and more}

18: % \subclass{MSC code1 \and MSC code2 \and more}

19: \end{abstract}

20: