export_corrected_variance_file

Output file for prediction variance calculations

Specification

  • Alias: None

  • Arguments: STRING

  • Default: corrected model variance export to default filename

Child Keywords:

Required/Optional

Description of Group

Dakota Keyword

Dakota Keyword Description

Optional (Choose One)

Tabular Format

custom_annotated

Selects custom-annotated tabular file format

annotated

Selects annotated tabular file format

freeform

Selects freeform file format

Description

Enables the user to specify the name and format of the file to which corrected model variance information is output. The file will contain the values of the variances computed at those points specified with num_predictions, prediction_configs, or import_prediction_configs.

For scalar responses, the prediction variance for each observable \(d_i\) is calculated according to

\[\Sigma_{total}(x) = \Sigma_{\delta}(x) + \sigma^2_{exp}I.\]

Here, \(\Sigma_{\delta}(x)\) is the (co)variance calculated from the Gaussian process or polynomial regression model, details of which can be found in the Dakota User’s and Theory Manuals. If experimental variance information is provided for the calibration, the maximum variance found for the observable \(d_i\) is used as \(\sigma^2_{exp}\) . For more details, see the Dakota User’s and Theory Manuals. For field responses, the variance of the discrepancy model alone is output to this file.

Default Behavior

If export_corrected_variance_file is not specified, information will be automatically output to dakota_discrepancy_variance_tabular.dat.

Usage Tips

Dakota exports tabular data in one of three formats:

  • annotated (default)

  • custom_annotated

  • freeform