Dynamic programming and stochastic control (Q800282): Difference between revisions

The basic objective of the book is to provide a unified framework for sequential decision making under uncertainty and to stress the few fundamental concepts underlying the treatment of uncertainty and the technique of dynamic programming. These concepts (risk, feedback, sufficient statistics, adaptivity, contraction mappings, and the principle of optimality) are emphasized and developed in a framework that is devoid, to the extent possible, of structural assumptions on the problem considered. This is accomplished by considering general dynamic systems defined over arbitrary state and control spaces. Thus our formulation allows the simultaneous treatment of several important classes of problems, such as stochastic control problems (popular in modern control theory) as well as problems of control of finite state Markov chains (popular in operations research and statistics).