On the Optimal Control of Partially Observed Inventory Systems

摘要: This paper introduces recent developments in the analysis of inventory systems with partial observations. The states these are typically conditional distributions, which evolve infinite dimensional spaces over time. Our involves introducing unnormalized probabilities to transform nonlinear state transition equations linear ones. With equations, existence optimal feedback policies proved for two models where demand and partially observed. In a third model current is not observed but past level fully observed, sufficient statistic provided serve as state. last serves an example has finite that model, we also establish optimality basestock policies, hence generalizing corresponding classical full information.