Convergence of stochastic approximation coupled with perturbation analysis in a class of manufacturing flow control models (Q1314853)

Stochastic approximation methods are examined for the solution of certain types of manufacturing flow control models having feedback control policies indexed over a compact finite-dimensional set \(\Theta\), where the objective functions are the average costs over a certain finite time interval. Representations are derived for the stochastic gradient of the objective function with respect to the control parameter \(\theta\in\Theta\), furthermore, conditions for the a.s. convergence of the stochastic gradient method towards the optimum are given.