evaluate_scenarios_parallel: Evaluate different scenarios in parallel

Description

This function takes a list of lists of scores, a list of lists of estimators, and lists lists of various other design parameters. Each possible combination of the elements of the respective sublists is then used to create separate scenarios. These scenarios are than evaluated independelty of each other, allowing for parallelization via the future framework. For each scenario, one call to the evaluate_estimator function is made.

Usage

evaluate_scenarios_parallel(
  score_lists,
  estimator_lists,
  data_distribution_lists,
  use_full_twoarm_sampling_distribution_lists,
  design_lists,
  true_parameter_lists,
  mu_lists,
  sigma_lists,
  tol_lists,
  maxEval_lists,
  absError_lists,
  exact_lists,
  early_futility_part_lists,
  continuation_part_lists,
  early_efficacy_part_lists,
  conditional_integral_lists
)

Value

a list of data.frames containing the results for the respective scenarios.

Arguments

score_lists: a list of lists of estimator scores.
estimator_lists: a list of lists of estimators.
data_distribution_lists: a list of lists of data distributions.
use_full_twoarm_sampling_distribution_lists: a list of lists of use_full_twoarm_sampling_distribution_lists parameters.
design_lists: a list of lists of designs.
true_parameter_lists: a list of lists of true parameters.
mu_lists: a list of lists of mu vectors.
sigma_lists: a list of lists of sigma values.
tol_lists: a list of lists of relative tolerances.
maxEval_lists: a list of lists of maxEval boundaries.
absError_lists: a list of lists of absError boundaries.
exact_lists: a list of lists of `exact` parameters.
early_futility_part_lists: a list of lists of `early_futility_part_lists` parameters.
continuation_part_lists: a list of lists of `continuation_part_lists` parameters.
early_efficacy_part_lists: a list of lists of `early_efficacy_part_lists` parameters.
conditional_integral_lists: a list of lists of `conditional_integral_lists` parameters.

Details

Concretely, the cross product of the first sublist of scores and the first sublist of estimators and the other parameters is calculated. Then the cross product of the second sublist of scores, estimators and other design parameters is calculated. All of these cross products together make up the set of all scenarios. The combinations say the first sublist of scores and the second sublist of estimators are not considered.

Examples

Run this code

res <-evaluate_scenarios_parallel(
 score_lists = list(c(MSE(), OverestimationProbability())),
 estimator_lists =  list(c(SampleMean(), FirstStageSampleMean())),
 data_distribution_lists = list(c(Normal(FALSE), Normal(TRUE))),
 design_lists =  list(c(get_example_design())),
 mu_lists = list(c(-1, 0, 1)),
 sigma_lists = list(1)
)