Abstract:
The convergence and rate of convergence of the Narendra-Shapiro automaton algorithm which rearranges the probabilities of choice of options for minimizing mean losses are investigated. The optimal parameters of the algorithm are determined.