No label defined (Q4693133)

[For the Russian original edition see Zbl 0604.60082]. The book deals with problems of mean value representations and numerical approximations to solutions of mainly classical initial-boundary value problems for the equations of mathematical physics via Monte-Carlo methods. The principal tool here consists of both mean values of paths and Wiener integral representations \(u=\underline E [I(\xi)]\) of the solutions through an appropriate random, in general branching and breaking-off, Markov process \(\xi\). These well known mean value theorems, discovered firstly for boundary value problems to Laplace operators and Brownian motions by N. Wiener, R. Courant, K. Friedrichs and H. Lévy, and developing for ever and a day, lead to further questions of computer simulations and approximations. The following subjects are considered: classification of computer simulated Markov processes arising from the mean value representations; rigorous construction of realizable unbiased estimators to u over paths \((\xi_ t)\) of random size \(\tau\) and their time-restricted analogues; and iterative processes for approximating solutions. Boundary value problems for linear and square-non-linear elliptic operators, parabolic operators, Lyapunov-Schmidt, second order Fredholm, Boltzmann type, Schrödinger and some other equations are considered. The 7 th and last chapter, added here to the Russian version of the book, presents a selected review of recent developments, related to the subjects: the propagation of chaos theory; the martingale method for many-particle systems and measure-valued diffusions [mainly by \textit{K. Oelschläger}, see Z. Wahrscheinlichkeitstheor. Verw. Geb. 69, 279-322 (1985; Zbl 0549.60071), and Ann. Probab. 12, 458-479 (1984; Zbl 0544.60097)]; numerical solutions and estimation of functionals for evolution SDEq's via the Neumann-Ulam scheme [see \textit{W. Wagner}, J. Comput. Phys. 71, 21-33 (1987; Zbl 0619.65011)]. The authors discuss open problems of modern mean value integral representation theory and Monte- Carlo methods.