Algorithm 425: generation of random correlated normal variables [G5]
We have programmed and made timing comparisons for two algorithms which sample the multivariate normal density N(&mgr;, V) = |V-1|/(2&pgr;)n/2·exp(- 1/2(Y - &mgr;)T V-1(Y - &mgr;)) (1) where V is an n X n covariance matrix, &mgr; is an n component vector of means, and Y is an n component random vector [1].