This research is concerned with two problems in sequential stochastic decision processes: the analysis of Markov Decision Processes with Reward/Variability tradeoffs and the computation of solutions in zero-sum stochastic games. Markov Decision Processes constitute a class of paradigms widely applicable to such areas as manufacturing systems, discrete stochastic control, etc. The Reward/Variability tradeoff approach holds the possibility of enhancing the acceptance of these paradigms by practitioners. The solution of zero-sum stochastic games is generally difficult; it is expected that the proposed approach will yield efficient algorithms for solving such problems.
