Simulate data from a fitted Rceattle model

The stats::simulate() method for CEATTLE fits: draws nsim replicate data sets from the fitted observation model, and optionally from the process model as well.

Usage

# S3 method for class 'Rceattle'
simulate(object, nsim = 1, seed = NULL, process = FALSE, ...)

Arguments

object

An object of class "Rceattle" returned by fit_mod().

nsim

Number of replicate data sets to draw. Default 1.

seed

Optional seed. The caller's random state is restored afterwards, per stats::simulate(). Recorded as attr(, "seed").

process

Which process error to redraw alongside the observations. FALSE (default) keeps the fitted deviations, so replicates differ only in observation error. See sim_mod() for the alternatives.

...

Currently unused.

Value

A list of nsim data_list objects, always a list, including at nsim = 1, so callers do not have to special-case the length. When process redrew something, each element holds the deviations that generated it as attr(, "process_sim"): a named list of whichever of rec_dev, init_dev, log_M1_dev and beta_linkage_re were drawn, each with a _drawn logical of the same shape marking the cells the draw touched. Compare estimates against those rather than against object, see sim_mod().

Details

A wrapper on sim_mod(), which documents the observation model in full. For expected values rather than draws, call sim_mod(simulate = FALSE).

Draws are taken by the TMB model, so simulate() needs a live $obj: a model loaded from disk has one, a model_average() result does not, see sim_mod().

See also

sim_mod() for the observation model and the process options, self_test() for simulating and refitting in one step.

Examples

if (FALSE) { # \dontrun{
fit  <- fit_mod(data_list = BS2017SS, estimateMode = 1)
reps <- simulate(fit, nsim = 10, seed = 1)
# ...and with recruitment redrawn as well:
reps <- simulate(fit, nsim = 10, seed = 1, process = "recruitment")
truth <- attr(reps[[1]], "process_sim")$rec_dev
} # }