R/rasym.r

In pbreheny/breheny: Miscellaneous Functions

#' Generate asymptotic sequence of random normal sums
#'
#' Returns the sum of bernoulli variables with different probabilities.
#'
#' @param n      Number of random variates to generate.
#' @param N      Maximum sample size.
#' @param len    Length of sequence; default: all
#' @param mu     Mean of normal distribution; default: 0
#' @param mean   Return means (instead of sums?); default: FALSE
#'
#' @examples
#' rasym(5, 100, 5)
#' rasym(7, 100, 5, mean=TRUE)
#'
#' @export
#'
#'

rasym <- function(n, N, len=N, mu=0, mean=FALSE) {
  if (N %% len != 0) stop('N is not an even multiple of length', call.=FALSE)
  gap <- floor(N/len) # Should warn?
  nn <- seq(gap, N, gap)
  Y <- matrix(rnorm(n*len, sd=sqrt(gap)), len, n) + gap*mu
  if (mean) {
    Z <- apply(Y, 2, cumsum)/nn
  } else {
    Z <- apply(Y, 2, cumsum)
  }
  dimnames(Z) <- list(nn, 1:n)
  Z
}