Set Up a Dakota Study

There are two ways to create a Dakota study in the Dakota GUI.  You may import an existing Dakota study or create a new Dakota study from scratch.

Import Dakota Study

To import a Dakota study, click on the Import Dakota Study link in the settings editor view for your project:

DakotaGuiTutorialImage17.png

Using the file dialog that pops up, find the Dakota study that you would like to import from your file system and click Open.  If you are following along with the Cantilever tutorial, an example Dakota input file has been provided:  dakota_cantilever_center.in

You should see it added to the table:

DakotaGuiTutorialImage18.png

You can edit the Dakota study by clicking on the link here (or by clicking on the study in the navigator tree) and then clicking the “Open input file” link in the new settings editor view:

DakotaGuiTutorialImage19.png

You should see the Dakota input file appear in your text editor:

DakotaGuiTutorialImage46.png

New Dakota Study

Alternately, if you wish to create a Dakota study from scratch, you can use the Dakota Study Wizard.  Click on the New Dakota Study link in the settings editor view for your project:

DakotaGuiTutorialImage20.png

If you have more than one simulation model in your project, a dialog will appear asking you to select an existing simulation model. (Note that for now, this wizard does not support creating a Dakota study in isolation from a simulation model)

DakotaGuiTutorialImage21.png

After selecting a simulation model, you will be presented with the wizard:

DakotaGuiTutorialImage22.png

You may use this wizard to define a Dakota study according to parameters and responses from your simulation model.  You can also create a blank study if you wish.  The controls on this wizard page are as follows:

  • Study Name – The name of your study.
  • Input File Name – The name of your input file.
  • Study Goal / Method Options – Dropdowns to help you narrow your focus on what type of Dakota study you want to perform.

After making your selections, click Next.

DakotaGuiTutorialImage23.png

The Identify Dakota Variables page lets you select which Dakota variables you want in your study by selecting from a list of available simulation model parameters.  Check the checkboxes of variables you want to use, and choose the appropriate variable type for each.  When you’re finished, your page might look like this:

DakotaGuiTutorialImage24.png

When you’re ready to proceed, click Next.

DakotaGuiTutorialImage25.png

On the Characterize Dakota Variables page, you can define lower bounds, upper bounds, etc. for your chosen variables.  Click Next when you’re ready to proceed.

DakotaGuiTutorialImage26.png

On the Response Functions page, choose which simulation model responses you want to be represented by your Dakota study.  Click Next when you’re ready to proceed.

DakotaGuiTutorialImage47.png

On the final page of the wizard, you can specify the number of concurrent runs (i.e. how many threads Dakota will use) and the name of the driver script you want to use.  If you haven't created or imported a driver script yet, you can auto-generate your own Python driver script by leaving the checkbox checked.  This will save you a lot of time, as all the links between Dakota study and simulation model are automatically scripted for you based on information you've provided in the wizard.  In fact, if you do this, you may skip the entire next page of the tutorial!

When you're finished, click Finish.

Editing Dakota Studies

Once you have your study in the Dakota GUI, you can edit the input file using either the text editor on the right, or settings editor controls that populate for every Dakota keyword.  For instance, you may click on a top-level keyword (such as environment) and you will see the following controls populate in the settings editor:

DakotaGuiTutorialImage27.png

From here, you can edit every child keyword that is available for your environment block.  For more information on the specific usage of Dakota keywords, see the Dakota reference manual.