A method for generating normal pseudorandom numbers. (Russian) Zbl 0614.65004

Let \(x=x_ 1,x_ 2,...,x_ n\) be independent random variables uniformly distributed on \(<-\sqrt{3},\sqrt{3}>\). Let \(A=\| a_{ij}\|\) be the orthogonal matrix of the n-dimensional Walsh transformation. Then AX is asymptotically (for \(n\to \infty)\) a vector of n independent normal random variables. This fact is proposed to be used for the generation of normal random variables. It states that the proposed method offers for great n a quicker generation of normal random variables on computers. No empirical evidence and/or proofs of this assertion is given. Similar arguments should hold for the Fourier transformation.
Reviewer: J.Král


65C10 Random number generation in numerical analysis