allocation_control

Sample allocation approach for multilevel expansions

Specification

  • Alias: None

  • Arguments: None

Child Keywords:

Required/Optional

Description of Group

Dakota Keyword

Dakota Keyword Description

Required (Choose One)

Multilevel Sample Allocation Control

estimator_variance

Variance of mean estimator within multilevel polynomial chaos

rip_sampling

Sample allocation based on restricted isometry property (RIP) within multilevel polynomial chaos

Description

Multilevel expansions, including regression-based polynomial chaos expansion (PCE) and function train (FT) expansions, require a sample allocation strategy. Three options are currently available:

  • allocation based on assuming a convergence rate for the estimator variance (for regression PCE)

  • restricted isometry property (RIP) sampling (for regression PCE via compressed sensing)

  • rank sampling (for FT)

Default Behavior

Current defaults, when allocation_control is not specified, are estimator variance for PCE and rank sampling for FT.