1: \begin{abstract}

2: We propose new restarting strategies for the accelerated coordinate descent method.

3: Our main contribution is to show that for a well chosen sequence of  restarting times, the restarted method has a nearly geometric rate

4: of convergence. A major feature of the method is that it can take profit of the local quadratic error bound of the objective function without knowing the actual

5: value of the error bound. 

6: We also show that under the more restrictive assumption that the objective function is strongly convex, any fixed restart period leads to a geometric rate of convergence.

7: Finally, we illustrate the properties of the algorithm on a regularized logistic regression problem and on a Lasso problem.

8: 

9: 

10: %\keywords{Accelerated coordinate descent \and restarting strategies \and unknown strong convexity \and local quadratic error bound}

11: \end{abstract}

12: