An optimum random-search algorithm is considered. The convergence conditions to the greatest increase (local properties) and convergence to the point of extremum (integral properties) of a function by optimizing in the presence of noise, are found. The results are used for finding a global...

An iterative algorithm is proposed for selecting the best system among a finite number of stable stochastic systems on the basis of a certain performance index. It is assumed that the stochastic processes which are associated with these systems are regenerative processes, whose characteristics...