1: \begin{abstract}

2: This paper considers the problem of minimizing an expectation function over a closed convex set, coupled with an expectation constraint on either

3: decision variables or problem parameters. We first present a new stochastic approximation (SA) type algorithm, namely the cooperative SA (CSA),

4: to handle problems with the expectation constraint on devision variables. We show that this algorithm exhibits

5: the optimal ${\cal O}(1/\sqrt{N})$ rate of convergence, in terms of both optimality gap and constraint violation,

6: when the objective and constraint functions are generally convex, where $N$ denotes the number of iterations. Moreover, we show that

7: this rate of convergence can be improved to

8: ${\cal O}(1/N)$ if the objective and constraint functions are strongly convex.

9: We then present a variant of CSA, namely the cooperative stochastic parameter approximation (CSPA) algorithm, to deal with

10: the situation when

11: the expectation constraint is defined over

12: problem parameters and show that it exhibits similar optimal rate of convergence to CSA.

13: It is worth noting that CSA and CSPA are primal methods  which do not require the iterations on the dual space and/or

14: the estimation on the size of the dual variables.

15: To the best of our knowledge, this is the first time that such optimal SA methods for

16: solving expectation constrained stochastic optimization are presented in the literature.

17: 

18: %Some promising numerical results for

19: %these algorithms applied to some portfolio optimization and machine learning problems.

20: 

21: 

22: \vspace{.1in}

23: 

24: \noindent {\bf Keywords:} convex programming, stochastic optimization, complexity,

25: subgradient method

26: 

27: \vspace{.07in}

28: 

29: \noindent {\bf AMS 2000 subject classification:} 90C25, 90C06, 90C22, 49M37

30: 

31: \end{abstract}

32: