pirouette (version 1.6.9)

select_experiments: Select the experiments to do a Bayesian inference with.

Description

Select the experiments to do a Bayesian inference with.

Usage

select_experiments(
  experiments = list(create_test_experiment()),
  marg_liks = NULL,
  verbose = FALSE
)

Value

a list of inference models

Arguments

experiments

a list of one or more pirouette experiments, as can be created by create_experiment. If more than one experiment is provided and a "generative" experiment is part of them, the "generative" one has to be the first in the list. See also:

• Use check_experiments to check the list of experiments for validity

• Use create_all_experiments to create experiments with all combinations of tree model, clock model and tree priors

• Use create_all_bd_experiments to create experiments with all combinations of tree model, clock model and tree priors, except for only using birth-death tree priors

• Use create_all_coal_experiments to create all experiments with all combinations of tree model, clock model and tree priors, except for only coalescent tree priors

• Use shorten_experiments to shorten the run time of the list of experiments

marg_liks

a data frame with marginal likelihoods/evidences. A test data frame can be created by create_test_marg_liks

verbose

if TRUE, show more output

Author

Richèl J.C. Bilderbeek

See Also

Use check_experiments to check if an object is a list of experiments

Examples

if (beautier::is_on_ci()) {

  evidences <- create_test_marg_liks(
    site_models = list(beautier::create_jc69_site_model()),
    clock_models = list(beautier::create_strict_clock_model()),
    tree_priors = list(
      beautier::create_yule_tree_prior(), 
      beautier::create_bd_tree_prior()
    )
  )
  evidences$weight <- c(0.9, 0.1) # in favor of Yule

  if (rappdirs::app_dir()$os != "win") {
    experiment_yule <- create_test_cand_experiment()
    experiment_bd <- create_test_cand_experiment()
    experiment_bd$inference_model$tree_prior <- 
      beautier::create_bd_tree_prior()
    experiment_yule$beast2_options <- experiment_bd$beast2_options
    experiment_yule$inference_model$mcmc <- experiment_bd$inference_model$mcmc
    experiment_yule$errors_filename <- experiment_bd$errors_filename
    experiments <- list(experiment_yule, experiment_bd)

    # Select the experiment.
    # In this case, select the candidate experiment with the highest evidence
    select_experiments(
      experiments = experiments,
      marg_liks = evidences
    )
  }

}