model_selection

Perform a recursion of admissible model subsets for a given model ensemble

Specification

  • Alias: None

  • Arguments: None

Description

This option specifies an enumerative search over all model subsets for a given model ensemble. In the MFMC case, there is a single DAG definition per model subset. Thus the model subset search is not combined with a DAG search for MFMC. The model subset with the best performance (lowest estimator variance for a prescribed budget or lowest cost for a prescribed accuracy) is selected for final post-processing.

Examples

method,
  model_pointer = 'ENSEMBLE'
  multifidelity_sampling
    search_model_graphs
      model_selection
  max_function_evaluations = 250
  pilot_samples =  25
  seed = 1237