1: \begin{abstract}

2: In \cite{Pang_Dist_Dyk,Pang_sub_Dyk}, we designed a distributed deterministic

3: asynchronous algorithm for minimizing the sum of subdifferentiable

4: and proximable functions and a regularizing quadratic on time-varying

5: graphs based on Dykstra's algorithm, or block coordinate dual ascent.

6: Each node in the distributed optimization problem is the sum of a

7: known regularizing quadratic and a function to be minimized. In this

8: paper, we prove sublinear convergence rates for the general algorithm,

9: and a linear rate of convergence if the function on each node is smooth

10: with Lipschitz gradient. Our numerical experiments also verify these

11: rates.

12: \end{abstract}

13: