Stochastic control processes and management science (Q1082328)

The author studies a formulation of stochastic control problems meant to describe situations in which the goals, consequences of and constraints on possible actions are imprecisely known. He argues that a useful formulation of this type should differentiate and account separately for random and ''fuzzy'' effects. The model considered is an ordinary differential equation whose right-hand-side is subject to random perturbations. In the random formulation the criterion of ''greater importance'' is said to be the probability that the solution stays above or below some level. In the fuzzy formulation the criterion (to be minimized) is a membership function (although an example of such a function is given, its definition is not). The paper's main (unproved) results concern existence and uniqueness of the optimal membership function and the form of an optimal control in the fuzzy formulation. The paper ends with brief descriptions of some possible applications of the results. The text of this work (''submitted by R. Bellman'') is not easy to follow. Its syntactical and other obscurities could have been removed by the editorial supervision that it apparently did not receive.