No label defined (Q4339077)

The book concerns not only the ``optimal control'' as it is suggested by its title but also the control theory understood in the wider sense of the control of any dynamic system (e.g., communication, distribution, production, financial, economic), in general stochastic and imperfectly observed. In connection with this the dynamic system models and related problems like controllability, observability, stability and feedback are also considered. The author takes this wider view and so covers general techniques of optimization based mainly on dynamic programming and the maximum principle, but also, e.g., on Lagrange multipliers. Both, discrete-time and continuous-time systems and related problems are considered in parallel over the whole book. The contents of the book is composed of two subtitles; the first of them, ``Basics'', contains the standard material concerning dynamic system models and optimization techniques; the second, ``Beyond'', contains some more advanced problems related with recent results. The material of the subtitle ``Basics'' occupying over 60\% of the volume is composed of two parts. Part 1 entitled ``Deterministic Models'' is successively described in Chapters 2-7 concerning the following problems. Deterministic models and their optimization; An infinite horizon behaviour; The classic formulation of the control problem: operators, filters, feedback; State space deterministic models: controllability, observability stability; Stationary rules and direct optimization of the LQ model; The Pontryagin maximum principle. Part 2 entitled ``Stochastic Models'' described in Chapters 8-15 contains the following problems. Stochastic dynamic programming; Stochastic dynamics in continuous time; Some stochastic examples; Policy improvement: stochastic versions; The LQG model with imperfect information; Stationary processes -- spectral theory; Optimal allocation; Imperfect state observation. The second subtitle ``Beyond'' is divided into further three parts. Part 3 entitled ``Risk-sensitive and \(H_\infty\) Criteria'' contains Chapters 16 and 17 with the problems: Risk-sensitivity; The \(H_\infty\) formulation. Part 4 entitled ``Time-integral Methods and Optimal Stationary Policies'' is composed of Chapters 18-21 with the following problems. The time-integral formalism; Optimal stationary LQG policies-perfect observation; Optimal stationary LQG policies-imperfect observation; The risk sensitive version. Part 5 entitled ``Near-determinism and Large Deviation Theory'' contains the Chapters 22-25 with the following problems. The essentials of large deviation theory; Control optimization in the large deviation limit; Controlled first passage; Imperfect observation -- nonlinear filtering. The advantage of the book is an original material presentation basing on many examples taken from different fields, which introduce the reader to not simple problems of optimal control. From the other side, the author avoids the derivations of the formulas and the shortcuts appearing in the considerations need proficiency of the reader in using basic mathematics. The results of considerations are usually presented in the form of theorems with explanations and without proofs which are only outlined. The advantage of this approach consists on a compact presentation of the main results. However, the reader may be not convinced that he is able to prove the theorems. Reading this book need some appropriate preparation.