On the speed of convergence of the kiefer-wolfowitz stochastic approximation procedure

In this article the speed of convergence of the Kiefer-wolfowitz procedure of stochastic approximation is scudied. This is done in Section 2 for two classes of regression functions in the unconstrained case. In Section 3 constrained problems are examined, on the one hand by transforming them into unconstrained ones and making use of the results of Section 2 and on the other hand by applying a projection method. Finally a problem from economic theory is formalized as a stochastic approximation problem and a solution is given oy means of the proceding theorems.