Dakota Reference Manual  Version 6.4
Large-Scale Engineering Optimization and Uncertainty Analysis
 All Pages

Selects annotated tabular file format


This keyword is related to the topics:


Alias: none

Argument(s): none

Default: annotated format


An annotated tabular file is a whitespace-separated text file with one leading header row of comments/column labels. Each subsequent row contains an evaluation ID and interface ID, followed by data for variables, or variables followed by responses, depending on context.

Default Behavior

By default, Dakota imports and exports tabular files in annotated format. The annotated keyword can be used to explicitly specify this.

Usage Tips

  • To specify pre-Dakota 6.1 tabular format, which did not include interface_id, specify custom_annotated header eval_id
  • Prior to October 2011, calibration and surrogate data files were in free-form format. They now default to annotated format, though freeform remains an option.
  • When importing tabular data, a warning will be generated if a specific number of data are expected, but extra is found and an error generated when there is insufficient data.
  • Some TPLs like SCOLIB and JEGA manage their own file I/O and only support the freeform option.


Export an annotated top-level tabular data file containing a header row, leading eval_id and interface_id columns, and data for variables and responses. Input file fragment:

    tabular_data_file = 'dakota_summary.dat'

Resulting tabular file:

%eval_id interface             x1             x2         obj_fn nln_ineq_con_1 nln_ineq_con_2 
1            NO_ID            0.9            1.1         0.0002           0.26           0.76 
2            NO_ID        0.90009            1.1 0.0001996404857   0.2601620081       0.759955 
3            NO_ID        0.89991            1.1 0.0002003604863   0.2598380081       0.760045