Constrained optimization via stochastic approximation with a simultaneous perturbation gradient approximation · Study