A method of solving a stochastic programming problem with constraints which are almost surely satisfied (Q5947821)

A convex problem of stochastic programming with restrictions which almost surely will be satisfied, i.e. the number of restrictions is equal to infinity, is considered. The notion ``almost surely satisfied'' is understood with respect to a measure of a probability space. To solve the problem a stochastic algorithm with use of a regularizing method of restrictions aggregation is introduced. The convergence of the algorithm at the average in the non-regular case, and its divergence with the probability equal to 1 in the regular case -- are proved.