On the convexity of the value function in Bayesian optimal control problems (Q1341493)

The principal purpose of the paper is to give proofs accessible to economists of Blackwell's theorem from 1951. The author studies the convexity of the value function, Blackwell's theorem and relates this to the uniqueness of optimal policies. The strict convexity and the strict inequality in Blackwell's theorem hold if and only if from different priors, different optimal actions may be chosen.