Skip to contents

run_simulation_benchmark() still returns a single list for tests. Scripts persist four RDS files keyed by ID so the metrics object does not duplicate geometry, MixSim, or \theta.

Usage

write_simulation_artefacts(
  benchmark,
  dir,
  stem,
  config = NULL,
  rewrite_bivariate_id = FALSE
)

Arguments

benchmark

List from run_simulation_benchmark().

dir

Output directory.

stem

File-name stem (bivariate, hybrid, …).

config

Optional scenario grid (defaults to benchmark$config).

rewrite_bivariate_id

If TRUE, rebuild fig02-style IDs.

Value

Invisibly, a named list of written paths.