Check a prior against a model — validate

brms's validate_prior(): resolve a prior specification against the model it is meant for, refuse it if any part of it addresses nothing the model has, and return the whole default_prior() table with the prior filled in. Fitting runs the same check, so a prior this function accepts is one frm() accepts, and a prior it refuses is refused here with the message the fit would give, before any fitting work.

Usage

validate_prior(
  prior,
  formula,
  data,
  family = NULL,
  data2 = list(),
  route = c("fit", "sample")
)

Arguments

prior: A frmtmb_priorlist from set_prior(), a brmsprior, or NULL for none.
formula: A bf() formula (with family) or a plain formula.
data: A data frame of model data (ignored when object is a fit).
family: Family, when object does not carry one.
data2: Structural objects, as in frm() (ignored when object is a fit, which carries its own).
route: Which route's defaults the prior column reports. "fit" (the default) reports the defaults frm() applies, which are flat in every slot; it consults no registry, so its answer does not depend on which packages are attached. "sample" reports the defaults frmtmb.sample::frm_sample() applies, and refuses when no package has registered any. The returned object records the route and print() names it on its first line.

Value

A frmtmb_prior_rows table, as default_prior() returns.

Details

A row the prior sets reads source = "user". A row that takes its density from a class row above it, the coefficients under a class "b" prior or the blocks under a class "sd" prior, reads source = "(vectorized)" and repeats that density, as brms prints it. Every other row keeps its default.

Two specifications for the same slot are refused, as frm() and brms refuse them. Where specifications for different slots reach one parameter, the table shows the more specific one, which is the one frm() applies; see set_prior().

The table can be passed back as prior =: frm() reads a frmtmb_prior_rows table the way as.brmsprior() does, applying the rows with a density or a bound and skipping the flat and vectorized ones.

Examples

dd <- data.frame(y = rnorm(60), x = rnorm(60), z = rnorm(60),
                 g = factor(rep(1:6, 10)))
validate_prior(prior(normal(0, 10), class = b) +
                 prior(cauchy(0, 2), class = sd),
               y ~ x + z + (1 | g), data = dd)
#> route = "fit": the prior defaults frm() applies
#>           prior     class      coef group resp dpar nlpar lb ub       source
#> 1        (flat) Intercept                                 NA NA      default
#> 2 normal(0, 10)         b                                 NA NA         user
#> 3 normal(0, 10)         b         x                       NA NA (vectorized)
#> 4 normal(0, 10)         b         z                       NA NA (vectorized)
#> 5  cauchy(0, 2)        sd                                 NA NA         user
#> 6  cauchy(0, 2)        sd               g                 NA NA (vectorized)
#> 7  cauchy(0, 2)        sd Intercept     g                 NA NA (vectorized)
#> 8        (flat)     sigma                                 NA NA      default
# a prior on a coefficient the model does not have is refused
try(validate_prior(prior(normal(0, 1), coef = w), y ~ x, data = dd))
#> Error : Prior target not found (class=b, coef=w)