Number-theoretic methods in statistics

the long version

The application of number-theoretic methods is a new, but rapidly expanding, branch of statistics. The Monte Carlo method is already established, with wide applications in science and technology. In applying it, however, a set of 'pseudo' random numbers is required for statistical simulation, and the use of these numbers often leads to unacceptably large errors. The essence of the number-theoretic method described in this book is to reduce such errors by using number theory to find a set of points (sometimes called quasi random numbers) which can then be regarded as the representatives of a given distribution. The number-theoretic method is hence also known as the quasi or deterministic version of the Monte Carlo method. Number-theoretic Methods in Statistics gives the reader various methods of generating quasi random numbers and demonstrates their applications in solving a variety of statistical problems, for example, the numerical evaluation of probabilities and moments, optimization, experimental design including design of computer experiments and statistical inference.