Convergence and asymptotic agreement in distributed decision problems. (English) Zbl 0535.90006

The team decision problem is considered, and in particular the distributed problem in which each agent has an objective cost function and a prior probability distribution. In contrast to other schemes for distributed decision making (or computation) the authors are interested in consensus and in common decisions quite independently of implementation issues, where each agent does not specialize in updating some components of the decision vector assigned to him, but updates the entire decision vector. It is supposed that each agent obtains a different stochastic measurement possibly at different random times, which is related to the same uncertain random vector from the environment. Conditions for asymptotic convergence of each agent’s decision sequence and asymptotic agreement of all agents’ decisions are derived. The appealing and weak features of the suggested model are discussed.
Reviewer: F.V.Burshtein


91B10 Group preferences
91A35 Decision theory for games
90B50 Management decision making, including multiple objectives
Full Text: DOI