Where each declared distribution's arguments are computed in the model

rxEtaDistExpand() hoists every family argument onto its own model line, rxEdA.<eta>.<role>, so a declaration's arguments are computed per observation by the compiled model like any other model quantity – covariates included, through the ordinary covariate machinery, inside the ODE model pool. This says which model variable holds each argument, so an estimator can READ them from the solve instead of evaluating the argument expressions a second time in its own code.

Usage

rxUiEtaDistAnchors(ui, d = NULL)

Arguments

ui: rxode2 ui, model function, or solve-ready object – the EXPANDED model, since that is what carries the anchor lines
d: the declarations, as rxUiEtaDists() returns them (name and etaDist columns). Defaults to reading them off ui, which works before the expansion but NOT after: rxEtaDistExpand() removes etaDist from the iniDf, so an expanded ui reports no declarations and this would return an empty list. An estimator holds the declarations it stashed before expanding and should pass them.

Value

named list, one character vector per declared random effect, in argument order; empty list when nothing is declared

Details

The order is the declaration's ARGUMENT order, which rxEtaDistExpand() normalizes at the storage point, so element t lines up with argument t of the family.

NA means that argument has no anchor: a family whose quantile template never references it gets no model line for it (a normal-based family collapses to the latent). An estimator must fall back to its own value there, and a covariate in such an argument is invisible to the model – so NA alongside a covariate is a gap, not a value to work around.

Author

Matthew L. Fidler

Examples

# \donttest{
if (requireNamespace("lotri", quietly = TRUE) &&
      "lotriEtaDists" %in% getNamespaceExports("lotri")) {
mod <- function() {
  ini({
    lclm <- log(5)
    lclrv <- log(0.09)
    tv <- 3.45
    dist(eta.cl) ~ dgamma(shape = 1 / exp(lclrv),
                          rate = 1 / (exp(lclrv) * exp(lclm)))
    add.sd <- 0.7
  })
  model({
    cl <- eta.cl
    v <- exp(tv)
    linCmt() ~ add(add.sd)
  })
}
.d <- rxUiEtaDists(mod)
rxUiEtaDistAnchors(rxEtaDistExpand(mod(), param = "direct"), .d)
}
# }