R/buildScaleMat_function.R

In senresearch/lcest: Left Censored Estimates

#' Build Scale Matrix
#' 
#' This function builds a 2x2 scale matrix given xsd, ysd, corr, and df.
#' @param xsd A positive number representing the standard deviation of x.
#' @param ysd A positive number representing the standard deviation of x.
#' @param r  A number between -1 and 1 representing the correlation of x and y. 
#' @param df Integer greater than 3, representing degrees of freedom with df=Inf implying normal.
#' @return Returns a 2x2 matrix oftened used to give scaling parameters to rmt.
#' @export
#' @examples 
#' buildScaleMat( 1,1,0,Inf) #normal 
#' buildScaleMat( 1,1,0,4) #t with df =4
#'

buildScaleMat <- function(xsd, ysd, r, df){
  # force bounds
  if( xsd <= 0 || ysd <= 0){ stop("xSd and ySd must be positive.")}
  if (r > 1 || r < -1) {stop("r must be between or equal to -1 and 1.")}
  if (df < 3 ) {stop(" df must be greater than 3.")}
  if ( df %% 1 != 0 && df != Inf) {stop( "df must be integer.")}
  # get variance covaraince matrix
  varcovarMat <-matrix( c(xsd^2, xsd*ysd*r, xsd*ysd*r, ysd^2), nrow =2, ncol =2)
  # convert to scale matrix by calling convert to scale which multiplies by df based coeffeicent 
  scaleMat <- convertToScale( varcovarMat, df)
  scaleMat
}