Optimal decision processes and algorithms (Q922943)

The paper considers a decision model based on sequential information gathering actions. A ``true state'' is searched for in a linearly ordered state space or a Cartesian product of such spaces. Each experiment yields a subset of the state space which includes the true state. The approach is interpreted in terms of multiattribute decisions, of parallel computing, and of working in a group. Conditions are derived which are sufficient for using a dynamic programming approach to the construction of optimal algorithms or decision processes.