1: \begin{abstract}

2: This paper considers stochastic subgradient mirror-descent method

3: %, in combination with simple iterate averaging,

4: for solving constrained convex  minimization problems.

5: In particular, a stochastic subgradient mirror-descent method

6: with weighted iterate-averaging is investigated and its per-iterate convergence rate is analyzed.

7: The novel part of the approach is in the choice of weights that are used to construct the averages.

8: Through the use of  these weighted averages, we show that the known

9: optimal rates can be obtained with simpler algorithms than those currently existing in the literature.

10: Specifically, by suitably choosing the stepsize values, one can obtain the rate of the order

11: $1/k$ for strongly convex

12: functions, and the rate $1/\sqrt{k}$ for general convex functions (not necessarily differentiable).

13: Furthermore, for the latter case, it is shown that a stochastic subgradient mirror-descent

14: with iterate averaging

15: converges (along a subsequence) to an optimal solution, almost surely,

16: even with the stepsize of the form

17: $1/\sqrt{1+k}$, which was not previously known. The stepsize choices that achieve the best rates are

18: those proposed by Paul Tseng for acceleration of proximal gradient methods~\cite{Tseng2008}.

19: \end{abstract}

20: