"A Metaheuristic Approach for Correlated Random Vector Generation"
Edgard Mauricio Hurtado Medina, Oscar Guaje, Andrés L. Medaglia and Jorge Sefair

The generation of correlated random variables is relevant in the stochastic simulation of financial and manufacturing systems, among many other applications. The generally accepted techniques to generate correlated multivariate observations rely on the mathematical attributes of the probability density functions of the random variables. In this paper, we propose a new approach based on Iterative Local Search (ILS) that induces a desired correlation structure to multivariate random data independently of the probability density function of the input variables. The proposed methodology is able to improve the quality of the results found by the Iman & Conover method – currently used in commercial simulators such as Crystal Ball – at a low computational cost.