numerical_solve

Specify the situations where numerical optimization is used for MFMC sample allocation

Specification

  • Alias: None

  • Arguments: None

Child Keywords:

Required/Optional

Description of Group

Dakota Keyword

Dakota Keyword Description

Optional (Choose One)

Employ numerical solve

fallback

Fall back to a numerical solve when needed for mitigation in MFMC

override

Replace MFMC analytic allocation with a numerical solution

Optional (Choose One)

Optimization Solver

sqp

Use a sequential quadratic programming method for solving an optimization sub-problem

nip

Use a nonlinear interior point method for solving an optimization sub-problem

global_local

Use a hybrid global-local scheme for solving an optimization sub-problem

competed_local

Use a competed local solver scheme for solving an optimization sub-problem

Optional

solver_metric

Metric employed during numerical solutions in sampling-based multifidelity UQ methods.

Description

Multifidelity Monte Carlo (MFMC) supports an analytic solution for the allocation of samples per model instance based on response correlations and relative simulation cost. In some situations (over-estimated pilot sample, mis-ordered model correlations), this analytic solution may be either sub-optimal or undefined, requiring mitigation.

This specification allows for control of this mitigation; in particular, whether recourse to a numerical solution is strictly a fallback (default) or is desired as an unconditional override (regardless of the need for specific mitigations).

Further, when a numerical solve is employed, it can utilize either the sqp or nip solver options.

Default Behavior Analytic is preferred, with fallback to numerical only when mitigation is required.