The talk will discuss team games and person-by-person optimality for distributed dynamical decision system modeled by stochastic differential equations driven by Brownian motion, when the decision makes have different information structures. Necessary optimality conditions based on backward-forward stochastic differential equations and Hamiltonian functionals are presented, as well as, sufficient optimality conditions, for randomized and nonrandomized strategies. Examples of dynamical team games with decision makers having different information structures are presented to illustrate the application of minimum principle. Throughout the talk we illustrate how the existing theory based on classical centralized decision strategies should be modified to handle nonclassical decentralized strategies.
Group for Research in Decision Analysis