Theory of Selection Operators on Hyperspaces and Multivalued Stochastic Processes

In this paper, a new concept of selection operators on hyperspaces (subsets spaces) is introduced, and the existence theorems for several kinds of selection operators are proved. Using the methods of selection operators, we give a selection characterization of identically distributed multivalued random variables and completely solve the vector-valued selection problem for sequences of multivalued random variables converging in distribution. The regular selections and Markov selections for multivalued stochastic processes are studied, and a discretization theorem for multivalued Markov processes is established. A theorem on the asymptotic martingale selections for compact and convex multivalued asymptotic martingale is proved.