Stochastic differential equations, backward SDEs, partial differential equations (Q2442149): Difference between revisions

In 1990, the pioneering work by \textit{S. G. Peng} and the first author [Syst. Control Lett. 14, No.~1, 55--61 (1990; Zbl 0692.93064)] on nonlinear backward stochastic differential equations was published. It is a new class of stochastic differential equations with adapted solutions, the terminal value of which is given, and it can be understood as a generalisation of the martingale representation property which is combined with a stochastic differential equation. Backward stochastic differential equations had been studied already earlier, in particular by \textit{J.-M. Bismut} [J. Math. Anal. Appl. 44, 383--404 (1973; Zbl 0276.93060)] before the work by Peng and Pardoux [loc. cit.], but only in its linear version, as the adjoint equation in Pontryagin's maximum principle for stochastic control problems. The work on nonlinear backward stochastic differential equations by Peng and Pardoux [loc. cit.] has not only been at the origin of a new, quickly developing research area which has been attracting a lot of researchers of different countries and has been proving its dynamism and attractiveness with a regular large number of publications and regularly hold international conferences devoted to its topics. It has also been contributing by its vast field of applications to the development of other research topics, and with its ideas it has been at the origin of the development of more recent, new theories, in particular those of the so-called \(g\)-expectation and the related stochastic analysis on the nonlinear \(g\)-expectation space and its application in the theory of dynamic risk measures, but also the very recent theory of path-dependent partial differential equations. An important factor contributing to the attractiveness and the dynamic development of the theory of backward stochastic differential equations consists in its important applications in a large fields of different topics in mathematics and finance, among them option pricing and optimal portfolio strategies, stochastic control problems, stochastic differential games and the theory of second order partial differential equations and stochastic partial differential equations. The study of the relation between backward stochastic differential equations and partial differential equations described by the Feynman-Kac formula, which gives a stochastic interpretation of second order parabolic PDEs, and its extension to more general second order PDEs constitutes besides the study of the theory of backward stochastic differential equations the main objective of the textbook. Let us emphasise that both authors, Etienne Pardoux and Aurel Răşcanu, are both internationally recognised specialists in the concerned topic: Not only that {E. Pardoux} has been one of the both founders of the theory of backward stochastic differential equations in 1990, but with his following works together with \textit{S. Peng} [Lect. Notes Control Inf. Sci. 176, 200--217 (1992; Zbl 0766.60079)], he was also at the origin of the study of backward equations as generalisation of the Feynman-Kac formula to semi-linear parabolic and elliptic PDEs, and he has been working on generalisations of the Feynman-Kac formula and its application in different fields, namely that of homogenisation of PDEs. Aurel Răşcanu is a well-known specialist for multi-valued stochastic differential equations, and together with Etienne Pardoux he is the author of the pioneering papers on backward stochastic variational inequalities -- a generalisation of backward stochastic differential equations -- which have allowed to generalise the Feynman-Kac formula to parabolic variational inequalities and to give to them a stochastic interpretation. He also has made important contributions to the theory of viability for backward stochastic differential equations and related PDEs. The authors' monograph is a textbook, suitable for Master and PhD students but also for researchers with a minimal knowledge in stochastic analysis. It begins with a first chapter recalling the necessary background in stochastic analysis, among it conditional expectations, continuous martingales and Brownian motion -- all these notions are presented in a succinct manner and without proofs, but with a collection of exercises which are helpful for the following chapters. Afterwards, the textbook guides the reader from these basics via the chapters on the Itô stochastic calculus, on ordinary stochastic differential equations and the classical Feynman-Kac formula, but also on stochastic differential equations with multivalued drift -- to the best of my knowledge the first complete presentation of this topic in the given high quality in a textbook -- to the chapter on backward stochastic differential equations (backward equations with deterministic terminal time, with random terminal time and backward equations with subdifferential operator) and the corresponding generalisations of the classical Feynman-Kac formula to semilinear parabolic PDEs in the whole space, but also to those with Dirichlet and with Neumann boundary condition, to the corresponding elliptic PDEs and to parabolic variational inequalities. A section in this chapter is devoted to problems of viability for backward stochastic differential equations, i.e., for conditions which guarantee that the solution of an \(\mathbb{R}^n\)-valued backward equation lives in an a priori given closed subset of \(\mathbb{R}^n\). All the chapters are closed by a collection of exercises which help to develop the understanding of the reader and give complementary information going beyond the subjects of the book, like, for instance, the discussion of the notion of \(g\)-expectation -- a nonlinear expectation introduced by \textit{S. Peng} [in: Backward stochastic differential equations. Harlow: Longman. 141--159 (1997; Zbl 0892.60066); Lect. Notes Math. 1856, 165--253 (2004; Zbl 1127.91032)]. Let us also emphasise that the annexes of the textbook give additional technical results used in the book, like, for instance, a recall of the notion of viscosity solution and different results on uniqueness for viscosity solutions, but also different elements of nonlinear analysis, like, for instance, a recall of the notions of maximal monotone operators, compactness results, semiconvexity and semiconvex functions. All this makes that the present monograph gives a rather complete treatment of backward stochastic differential equations as tool for the stochastic interpretation of second order PDEs. As the reader is guided from basic knowledge on stochastic analysis through the Itō calculus and the theory of stochastic differential equations to that of the backward equations, the monograph represents in my eyes a precious textbook for Master students, PhD students, but also specialists in this domain.