1: \begin{abstract}

2: In this paper, we propose two novel non-stationary first-order primal-dual algorithms to solve nonsmooth composite convex optimization problems.

3: Unlike existing primal-dual schemes where the parameters are often fixed, our methods use pre-defined and dynamic sequences for parameters.

4: We prove that our first algorithm can achieve $\BigO{1/k}$ convergence rate on the primal-dual gap, and primal and dual objective residuals, where $k$ is the iteration counter. 

5: Our rate is on the non-ergodic (i.e., the last iterate) sequence of the primal problem and on the ergodic (i.e., the averaging) sequence of the dual problem, which we call semi-ergodic rate. 

6: By modifying the step-size update rule, this rate can be boosted even faster on the primal objective residual.

7: When the problem is strongly convex, we develop a second primal-dual algorithm that exhibits $\BigO{1/k^2}$ convergence rate on the same three types of guarantees.

8: Again by modifying the step-size update rule, this rate becomes faster on the primal objective residual.

9: Our primal-dual algorithms are the first ones to achieve such fast convergence rate guarantees under mild assumptions compared to existing works, to the best of our knowledge.

10: As byproducts, we apply our algorithms to solve constrained convex optimization problems and prove the same convergence rates on both the objective residuals and the feasibility violation.

11: We  still obtain at least $\BigO{1/k^2}$ rates even when the problem is ``semi-strongly'' convex.

12: We verify our theoretical results via two well-known numerical examples.

13: 

14: %%%% Keywords.

15: \vspace{1ex}

16: \noindent\textbf{Keywords:}

17: Non-stationary primal-dual method;  non-ergodic convergence rate; fast convergence rates; composite convex minimization; constrained convex optimization.

18: \end{abstract}

19: