Relationship between decentralized controller design using H/sub /spl infin// and stochastic risk-averse criteria

Considers discrete-time decentralized risk-averse LEQG (linear exponential of quadratic cost Gaussian) teams, and shows that the team-optimal solutions are identical to one of the solutions for minimax team problems. The author first gives the result for static teams, and extends the result to dynamic teams, with a quasi-classical information pattern, using a dynamic programming argument. In the process, the author also establishes differences between the two problems by pointing out the existence of multiple solutions to the minimax team problem, while there exists a unique solution to the LEQG team problem.<
>