R/transforms.R

In greta: Simple and Scalable Statistical Modelling in R

#' @name transforms
#' @aliases inverse-links
#' @title transformation functions for greta arrays
#'
#' @description transformations for greta arrays, which may also be used as
#'   inverse link functions. Also see [operators] and [functions].
#'
#' @param x a real-valued (i.e. values ranging from -Inf to Inf) greta array to
#'   transform to a constrained value
#' @details greta does not allow you to state the transformation/link on the
#'   left hand side of an assignment, as is common in the BUGS and STAN
#'   modelling languages. That's because the same syntax has a very different
#'   meaning in R, and can only be applied to objects that are already in
#'   existence. The inverse forms of the common link functions (prefixed with an
#'   'i') can be used instead.
#'
#'   The `log1pe` inverse link function is equivalent to `log(1 +
#'   exp(x))`, yielding a positive transformed parameter. Unlike the log
#'   transformation, this transformation is approximately linear for x > 1. i.e.
#'   when \eqn{x > 1}, \eqn{y} is approximately \eqn{x}
#'
#'   `imultilogit` expects an n-by-m greta array, and returns an n-by-(m+1)
#'   greta array of positive reals whose rows sum to one. This is equivalent
#'   adding a final column of 0s and then running the softmax function widely
#'   used in machine learning.
#'
#' @examples
#' \dontrun{
#'
#' x1 <- normal(1, 3, dim = 10)
#'
#' # transformation to the unit interval
#' p1 <- iprobit(x1)
#' p2 <- ilogit(x1)
#' p3 <- icloglog(x1)
#' p4 <- icauchit(x1)
#'
#' # and to positive reals
#' y <- log1pe(x1)
#'
#' # transform from 10x3 to 10x4, where rows are a complete set of
#' # probabilities
#' x2 <- normal(1, 3, dim = c(10, 3))
#' z <- imultilogit(x2)
#' }
NULL

#' @rdname transforms
#' @export
iprobit <- function(x) {
  op("iprobit", x,
    tf_operation = "tf_iprobit",
    representations = list(probit = x)
  )
}

#' @rdname transforms
#' @export
ilogit <- function(x) {
  op("ilogit", x,
    tf_operation = "tf$nn$sigmoid",
    representations = list(logit = x)
  )
}

#' @rdname transforms
#' @export
icloglog <- function(x) {
  op("icloglog", x, tf_operation = "tf_icloglog")
}

#' @rdname transforms
#' @export
icauchit <- function(x) {
  op("icauchit", x, tf_operation = "tf_icauchit")
}

#' @rdname transforms
#' @export
log1pe <- function(x) {
  op("log1pe", x, tf_operation = "tf$nn$softplus")
}

#' @rdname transforms
#' @export
imultilogit <- function(x) {
  dim <- dim(x)

  # check it's a matrix
  if (length(dim) != 2) {
    msg <- cli::format_error(
      c(
        "{.fun imultilogit} must have a 2D greta array",
        "{.code dim(x)} returns: {dim(x)}"
      )
    )
    stop(
      msg,
      call. = FALSE
    )
  }

  op("imultilogit", x,
    dim = dim + c(0, 1),
    tf_operation = "tf_imultilogit"
  )
}