Stochastic differential inclusions and applications. (Q1935213)

This book offers a comprehensive coverage of the theory of stochastic analysis for set-valued stochastic processes. The approach rests on characterizing set-valued mappings by a countable number of selectors to which a standard theory of (stochastic) integration applies. This involves intricate results about (weak) compactness and decomposability of set-valued functions. The book covers stochastic integrals of set-valued processes, stochastic integral/differential inclusions and their relation to partial differential inclusions. Last chapter is devoted to applications in stochastic control theory. Chapter 1 presents a theory of stochastic integration with respect to a Brownin motion and existence and uniqueness of solutions to stochastic differential equations. This is topped up with the introduction of an infinitesimal characterization of a process via generators and the Feynma-Kac formula. Elements of the general theory of stochastic processes are also discussed. Most of the results are provided with proofs. Chapter 2 introduces the theory of set-values mappings (multifunctions) with values in a Polish space. Results on the existence of measurable and continuous selectors are presented and a characterization of measurable multi-functions via a countable family of measurable selectors is established. These are later used for the construction of the Aumann integral. Set-valued random variables and stochastic processes are presented as a particular example of the theory. In particular, the conditional expectation is defined and its properties are proved. Chapter 3 presents a stochastic integration theory for set-valued processes. Similarly as for the Aumann integral, the construction of the stochastic integral (with respect to a Brownian motion) is based on Itô integration of previsible processes which are selectors of a set-valued stochastic process. Related results for the conditional expectation of stochastic integrals as well as the properties of set-valued Itô processes are proved. Chapter 4 deals with stochastic inclusions. A comprehensive theory is developed for stochastic functional inclusions (in particular, the existence of solutions). The main idea is to choose sufficiently regular selectors and then apply standard existence results for stochastic differential equations. Following this, the author introduces stochastic differential inclusions and shows the relationship between these and the functional inclusions. In Section 3, a special case (without the stochastic integral) of backward stochastic inclusions is studied. Chapter 5 extends results of the previous chapter to stochastic functional inclusions and backward stochastic differential inclusions with an additional constraint that the solution must be a selector of a fixed set-valued stochastic process. As a simple example one might think of a (standard) stochastic differential equation with the requirement that the solution lies in a random and time dependent set (in particular, in a fixed deterministic set). Main developments in this chapter lead towards establishing the existence of solutions. Chapter 6 looks at partial differential inclusions and establishes links between their solutions and solutions to stochastic functional inclusions in a similar fashion as the link between diffusions and partial differential equations. Chapter 7 looks at an optimal control problem of a stochastic differential equation. Standard approach for solutions leads through HJB equation and results in optimal control within the set of all sufficiently integrable previsible controls. The theory presented in the book enables optimization within other classes, e.g., deterministic controls or feedback controls specified by Lipschitz continuous functions. Such control problems are related to stochastic functional inclusions, where the drift and the diffusion coefficient are described as sets containing values corresponding to all possible controls. These are further linked to partial differential inclusions.