A strong approximation theorem for stochastic recursive algorithms

The constant stepsize analog of Gelfand-Mitter type discrete-time stochasti c recursive algorithms is shown to track an associated stochastic different ial equation in the strong sense, i.e., with respect to an appropriate dive rgence measure.