Abstract:
The paper deals with a special method of simulation of random vectors with specified identical one-dimensional marginal distributions and a specified positive correlation matrix. The main idea of the method is sampling of identical random variables for each group of a random subdivision of the set of components. A simple (but not always optimal) algorithm is introduced for the determination of suitable probabilities of different subdivision variants.
It is shown that the reproduction method is suitable for simulation of a stationary process with an absolutely monotone correlation function.