1: \begin{abstract}

2: We study connections between Dykstra's algorithm for projecting onto

3: an intersection of convex sets, the augmented Lagrangian method of

4: multipliers or ADMM, and block coordinate descent.  We prove that  

5: coordinate descent for a regularized regression problem, in which the

6: (separable) penalty functions are seminorms, is exactly equivalent to   

7: Dykstra's algorithm applied to the dual problem.  ADMM on the dual

8: problem is also seen to be equivalent, in the special case of two

9: sets, with one 

10: being a linear subspace. These connections, aside from being

11: interesting in their own right, suggest new ways of analyzing and

12: extending coordinate descent. For example, from existing convergence

13: theory on Dykstra's algorithm over polyhedra, we discern that

14: coordinate descent for the lasso problem converges at an

15: (asymptotically) linear rate.  We also develop two parallel versions 

16: of coordinate descent, based on the Dykstra and ADMM connections.        

17: \end{abstract}

18: