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

Launch analysis drivers using fork command


Alias: none

Argument(s): none

Required/Optional Description of Group Dakota Keyword Dakota Keyword Description
Optional parameters_file Specify the name of the parameters file
Optional results_file Specify the name of the results file
Optional allow_existing_results

Change how Dakota deals with existing results files

Optional verbatim

Specify the command Dakota uses to launch analysis driver(s) and filters

Optional aprepro Write parameters files in APREPRO syntax
Optional labeled Requires correct function value labels in results file
Optional file_tag

Tag each parameters & results file name with the function evaluation number

Optional file_save Keep the parameters & results files after the analysis driver completes
Optional work_directory

Perform each function evaluation in a separate working directory


The fork interface is the most common means by which Dakota launches a separate application analysis process.

The fork interface is recommended over system for most analysis drivers that are external to Dakota, i.e., any driver not linked in via the direct interface.

As explained in the Users Manual, the parameters and results file names are passed on the command line to the analysis driver(s). If input/output filters are specified, they will be run before/after the analysis drivers. The verbatim keyword is used to modify the default driver/filter commands.

For additional information on invocation syntax, see the Interfaces chapter of the Users Manual[5].


Spawn (fork) an external executable/script called 'rosenbrock' which reads variables from params.in and writes responses to results.out. Preserve the analysis files for each function evaluation with tag and save.

  analysis_drivers = 'rosenbrock'
      parameters_file = 'params.in'
      results_file   = 'results.out'