Asymptotics of the remainder term in the central limit theorem in a Hilbert space (Q1071373): Difference between revisions

The author shows that for a Fréchet differentiable function f with three derivatives and with third derivative bounded and continuous the expectation of a sum of n i.i.d. vectors in Hilbert space with finite third absolute moment can be approximated up to an error of order \(0(n^{-1/2})\) by the expectation of the corresponding Gaussian random vector. This well known result which has been proved by \textit{V. I. Paulauskas} [Litov. Mat. Sb. 16, No.3, 103-121 (1976; Zbl 0353.60014)] and \textit{V. M. Zolotarev} [Teor. Veroyatn. Primen. 21, 741-758 (1976; Zbl 0378.60003)] is extended to the form of an explicit remainder term which depends on the modulus of continuity of the third derivative and the difference between the distribution of the random vector and the corresponding Gaussian random vector. For abstract Wiener (Banach) spaces similar estimates for differentiable functions defined on the dense Hilbert subspace are derived.