relaxation
For an online pilot mode, apply under-relaxation to the shared sample increments
Specification
Alias: None
Arguments: None
Default: no relaxation
Child Keywords:
Required/Optional |
Description of Group |
Dakota Keyword |
Dakota Keyword Description |
|---|---|---|---|
Required (Choose One) |
Relaxation factor |
For under-relaxation of shared sample increments, apply a sequence of factors, one per iteration |
|
For under-relaxation of shared sample increments, apply a fixed factor that is invariant with iteration |
|||
For under-relaxation of shared sample increments, apply a recursive factor on each iteration that advances the relaxation factor toward 1 |
Description
Multilevel / multifidelity sampling methods are adaptive UQ methods
that utilize a pilot sample to estimate an initial set of correlations
or variances, and then augment the pilot with additional sample
increments to optimally allocate resources. In the iterated
online_pilot solution mode, the initial pilot sample is updated with
additional shared samples in the direction of the optimal solution to
the resource allocation problem. When the initial correlation /
covariance approximations are inaccurate, this update can be overly
aggressive, such that it becomes advisable to only accept a portion of
this suggested increment through use of an under-relaxation factor.
Three options are provided for defining a sequence of under-relaxation
factors: factor_sequence, fixed_factor, and recursive_factor.
Default Behavior
Relaxation is an optional specification, and no relaxation (accept the full magnitude of suggested sample increments) is the default.
