1: \begin{abstract}

2: In this paper, we develop a unified convergence analysis framework for the \textit{Accelerated Smoothed GAp ReDuction algorithm} (ASGARD) introduced in \cite[\textit{Tran-Dinh et al}]{TranDinh2015b}.

3: Unlike \cite{TranDinh2015b}, the new analysis covers three settings in a single algorithm: general convexity, strong convexity, and strong convexity and smoothness.

4: Moreover, we establish the convergence guarantees on three criteria: (i) gap function, (ii) primal objective residual, and (iii) dual objective residual.

5: Our convergence rates are optimal (up to a constant factor) in all cases.

6: While the convergence rate on the primal objective residual for the general convex case has been established in \cite{TranDinh2015b}, 

7: we prove additional convergence rates on the gap function and the dual objective residual.

8: The analysis for the last two cases is completely new.

9: Our results provide a complete picture on the convergence guarantees of ASGARD.

10: Finally, we present four different numerical experiments on a representative optimization model to verify our algorithm and compare it with Nesterov's smoothing technique. 

11: 

12: 

13: \keywords{

14: Accelerated smoothed gap reduction

15: \and

16: primal-dual algorithm

17: \and

18: Nesterov's smoothing technique 

19: \and

20: convex-concave saddle-point problem.

21: }

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

23: \subclass{90C25 \and 90C06 \and 90-08}

24: \end{abstract}

25: