R/beta.cauchy.R

In EbayesThresh: Empirical Bayes Thresholding and Related Methods

beta.cauchy <- function(x) {
#
#   Find the function beta for the mixed normal prior with Cauchy
#   tails.  It is assumed that the noise variance is equal to one.
#
	phix <- dnorm(x)
	j <- (x != 0)
	beta <- x
	beta[!j] <- -1/2
	beta[j] <- (dnorm(0)/phix[j] - 1)/x[j]^2 - 1
	return(beta)
}