Define Gradients

Gradients calculate the change of the output responses with respect to the input variables. The gradient information can be used in gradient enabled methods inside of a DOE, Fit, Optimization, or Stochastic approaches.

Gradients must be defined for all combinations of output responses and input variables; some or all gradients can remain undefined. An undefined gradient is not equivalent to setting a defined gradient to be equal to 0.

Gradients may also be defined using a combination of results read or extracted from a file and input variable values. These equations may use the functions available or external functions, including Compose/OML and Python functions. HyperStudy uses HyperWorks readers, and hence can read the binary output files of most popular CAE solvers. Additional data can be read using custom external readers and import templates. External readers, external functions and import templates may be manually added in the preference file or registered through the Edit menu.

  1. In the Define Output Responses step, click the Gradients tab.
  2. Click Add Gradient.
    A new gradient is created and added to the work area.
  3. In the Label column, enter a name for the gradient.
  4. In the Derivative of column, enter the varname of the output response to be associated with this gradient.
  5. In the Respect to column, enter the varname of the input variable to be associated with this gradient.
  6. Define the output response expression.
    • Manually enter the expression in the Expression column.
    • Build an expression using the Expression Builder, which can be accessed by clicking in the Expression column.
  7. Click Evaluate to extract the gradient value.
Tip: Automatically generate all possible gradient combinations by clicking Add All.