Efficiently randomly drawing from a multivariate normal distribution
Just wondering if anyone has ever encountered the problem where he/she needs to randomly draw from a very high dimensional multivariate normal distribution (say dimension = 10,000), as the rmvnorm function of the mvtnorm package is impractical for that. I know this article has an Rcpp implementation for the dmvnorm function of the mvtnorm package, so I was wondering if something equivalent exists for rmvnorm ? Here's a quick comparison of mvtnorm::rmvnorm and an Rcpp implementation given here by Ahmadou Dicko. The times presented are for 100 draws from a multivariate normal distribution with