R/stan-run-get_IVs.R

In banocc: Bayesian ANalysis Of Compositional Covariance

# Get initial values drawn from the prior
#
# @param chains The number of chains to run
# @param data A data object, which must be a list with elements:
#   \itemize{
#     \item \code{nu} The prior mean of mu
#     \item \code{Lambda} The prior variance/covariance matrix of mu
#     \item \code{eta} The parameter for the LKJ prior on the correlation
#       matrix
#     \item \code{alpha} The vector of shape parameters for the gamma prior on
#       sigma
#     \item \code{beta} The vector of scale parameters for the gamma prior on
#       sigma
#     \item \code{P} The number of features in the dataset
#   }
# @inheritParams cat_v
#
#' @importFrom mvtnorm rmvnorm

get_IVs <- function(chains, data, verbose=FALSE, num_level=0){
    cat_v("Begin get_IVs...", verbose, num_level=num_level)
    if (chains <= 0){
        stop("'chains' must be > 0")
    }
    IVs <- lapply(seq_len(chains),
                  function(i) {
                      list(m = as.vector(mvtnorm::rmvnorm(1, data$n,
                               data$L)),
                           O = diag(data$P),
                           lambda = rgamma(1, data$a, data$b))
                  })
    cat_v("Done.\n", verbose)
    return(IVs)
}