Abstract
We consider a dynamic finite-horizon multi-agent stochastic team problem. If the agents of the team do not share their observations with each other and do not recall their past observations, and the cost function of the team has a certain structure, then under certain technical conditions, the dynamic team is shown to admit a team-optimal solution in deterministic strategies of the agents. Consequently, we show that several dynamic LQG teams admit team-optimal solutions, and this class includes the celebrated Witsenhausen's counterexample, the Gaussian test channel, the Gaussian relay channel, and multi-dimensional versions of these dynamic teams.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
