R/J10.r

In logcondens: Estimate a Log-Concave Probability Density from Iid Observations

J10 <- function(x, y) {
  m <- length(x)
  z <- exp(x)
  d <- y - x
  II <- (1:m)[abs(d) > 0.01]
  z[II] <- z[II] * (exp(d[II]) - 1 - d[II]) / (d[II]^2)
  II <- (1:m)[abs(d) <= 0.01]
  z[II] <- z[II] *
    (1 / 2 +
      d[II] * (1 / 6 +
        d[II] * (1 / 24 +
          d[II] * (1 / 120 + d[II] / 720))))
  return(z)
}

# version prior to e-mails of LD of August 2010
# J10 <- function(x, y){
#     m <- length(x)
#     z <- exp(x)
#     d <- y - x
#     II <- (1:m)[abs(d) > 10 ^ (-6)]
#     z[II] <- z[II] * (exp(d[II]) - 1 - d[II]) / (d[II] ^ 2)
#     II <- (1:m)[abs(d) <= 10 ^ (-6)]
#     z[II] <- z[II] * (1 / 2 + d[II]/6 + d[II] ^ 2 / 24)
#     return(z)
# }