Stochastic dominance-constrained Markov decision processes

William B. Haskell, Rahul Jain

We are interested in risk constraints for infinite horizon discrete time Markov decision processes (MDPs). Starting with average reward MDPs, we show that increasing concave stocha…