R/models__linear__etc__delta.sumqlogdetK.R

In feng-li/movingknots: Efficient Bayesian multivariate surface regression

#' gradient for sum q_i log|K_i|
#'
#' Details
#' @param q.i NA
#' @param diag.K NA
#' @param args "list" args$subset: the subsets of the diagonal K matrix.
#' @return "matrix" 1-by-
#' @references NA
#' @author Feng Li, Department of Statistics, Stockholm University, Sweden.
#' @export
delta.sumqlogdetK <- function(q.i, diag.K)
  {
    p <- length(diag.K)/length(q.i)
    q.seq <- rep(q.i, each = p)
    ## subset <- args$subset
    out0 <- q.seq/diag.K # the gradient
    out <- matrix(out0, 1) # format to a matrix
    return(out)
  }