Dakota console scripts

These console scripts are called as Dakota’s analysis_driver: dakota_run_component for a CSDMS component, and dakota_run_plugin for a model interfaced by a Dakotathon plugin class.

The dakota_run_component script

Defines the dakota_run_component console script.

dakotathon.run_component.main()

Handle arguments to the dakota_run_component console script.

dakotathon.run_component.run_component(params_file, results_file)

Brokers communication between Dakota and a CSDMS component.

params_file : str

The path to the parameters file created by Dakota.

results_file : str

The path the results file returned to Dakota.

This console script provides a generic analysis driver for a Dakota experiment. At each evaluation step, Dakota calls this script with two arguments, the names of the parameters and results files:

1. The parameters file provides information on the current Dakota evaluation step, including the names and values of component variables and their responses. It also includes, as the analysis component, the name of a configuration file that stores information about the setup of the experiment, including the name of the component to call, input files, output file(s) to examine, and the statistic to apply to the output file(s).
2. The results file contains component output values in a format specified by the Dakota documentation.

Once the component is identified, a worker is created to perform three steps: preprocessing, execution, and postprocessing. In the preprocessing step, information from the configuration file is transferred to the component. In the execution step, the component is called, using the information passed from Dakota. In the postprocessing step, output from the component is read, and a single statistic (e.g., mean, median, max, etc.) is applied to it. This number, one for each response, is returned to Dakota through the results file, ending the Dakota evaluation step.

The dakota_run_plugin script

Defines the dakota_run_plugin console script.

dakotathon.run_plugin.main()

Handle arguments to the dakota_run_plugin console script.

dakotathon.run_plugin.run_plugin(params_file, results_file)

Brokers communication between Dakota and a model through files.

params_file : str

The path to the parameters file created by Dakota.

results_file : str

The path the results file returned to Dakota.

This console script provides a generic analysis driver for a Dakota experiment. At each evaluation step, Dakota calls this script with two arguments, the names of the parameters and results files:

1. The parameters file provides information on the current Dakota evaluation step, including the names and values of model variables and their responses. It also includes, as the analysis component, the name of a configuration file that stores information about the setup of the experiment, including the name of the model to call, input files, output file(s) to examine, and the statistic to apply to the output file(s).
2. The results file contains model output values in a format specified by the Dakota documentation.

Once the model is identified, an interface is created to perform three steps: preprocessing, execution, and postprocessing. In the preprocessing step, information from the configuration file is transferred to the component. In the execution step, the component is called, using the information passed from Dakota. In the postprocessing step, output from the component is read, and a single statistic (e.g., mean, median, max, etc.) is applied to it. This number, one for each response, is returned to Dakota through the results file, ending the Dakota evaluation step.