override

Replace MFMC analytic allocation with a numerical solution

Specification

  • Alias: None

  • Arguments: None

  • JSON argument key: None

  • Default: true

Description

Unconditionally employ a numerical solve for allocation of samples per model instance within multifidelity Monte Carlo (MFMC). The analytic allocation is replaced with numerical solution using either sqp or nip solvers.

Usage Tips An unconditional override can be appropriate in the case where you select a pilot sample that may be larger than the anticipated optimal sample profile results (pilot over-estimation). In this context, the numerical optimization has greater freedom to work around the incurred pilot cost and find a solution that optimally incorporates the pilot.

Additional Discussion

Examples

method
  model_pointer = 'ENSEMBLE'
  multifidelity_sampling
    numerical_solve nip override
    max_function_evaluations = 1000
    pilot_samples = 100
    seed = 8674132
{
  "method": [
    {
      "multifidelity_sampling": {
        "model_pointer": "ENSEMBLE",
        "numerical_solve": {"nip": {}, "override": {}},
        "max_function_evaluations": 1000,
        "pilot_samples": [100],
        "seed_sequence": [8674132]
      }
    }
  ]
}