Estimates for the deviations of random walks and a stochastic method for solving the Schrödinger equation. (Q556570)

The Cauchy problem for the Schrödinger equation on small time intervals arises in quantum chemistry, quantum optics, quantum theory of Brownian motions. The authors consider the stochastic representation of solution of the Cauchy problem for the Schrödinger equation with potential from the algebra \({\mathcal A}(\mathbb R^n)\) of the Fourier images of measures of bounded variation, and study problems which can be reduced to that case by using the interaction repesentation. The stochastic representation allows one to apply the Monte-Carlo algorithm [\textit{S. M. Ermakov} and \textit{G. A. Mikhajlov}, ``Statistical simulation'' (1982; Zbl 0599.65001)] for numerical estimates of the values of various physical observables. In Section 2, the authors derive explicit stochastic representations of the solution of the Cauchy problem for the Schrödinger equation cases reducible to the problem with potential from the class \({\mathcal A}(\mathbb R)\). The theory of the Monte-Carlo method for the Cauchy problem for Schrödinger equations with potentials from the algebra of the Fourier transformations of measures with bounded variation was described by \textit{A. M. Chebotarev} [Mat. Zametki 24, 699--706 (1978; Zbl 0402.60062)], \textit{A. M. Chebotarev} and \textit{V. P. Maslov} [in: Feynman path integrals. Lect. Notes Phys. 106, 58--72 (1979; Zbl 0443.28010)] and others. This method requires finite-difference approximations and in the momentum representation, it provides a simple estimate of the calculation error, depending only on the number of sampling trajectories of the Poisson process used for the approximation of the mathematical expectation. The numerical calculations of the test tunnelling problems for potentials which are superpositions of pair Gaussian functions were discussed by \textit{A. M. Chebotarev} and \textit{R. Quezada-Batalla} [Russ. J. Math. Phys. 4, No. 3, 275--286 (1996; Zbl 0911.60094)]. In the interaction representation, the method used by Chebotarev and Quezada-Batalla can be applied to potentials containing additional linear and quadratic polynomial components. In Section 3 the authors consider the derivation probabilities for random walks and derive analytical and numerical estimates. In Section 4, entitled `Unitary matrix approximation of the mathematical expectation by using the Cayley transformation', they construct the unitary matrix approximation of the resolving operator.