Configuring the Objectives

You can select the parameters that will be used as objectives during execution. The Approximation Loop and Pointer-2 strategies have the same objective attributes. In addition, because both strategies have the same general intent as the Optimization component, their design parameters have the same attributes as the Optimization component.

The Approximation Loop and Pointer-2 strategies are similar to a penalty-based single objective optimization algorithm. While optimization techniques used inside the strategies may be able to operate on the constraints directly, design improvements are evaluated by the strategies based on the single value of the ObjectiveandPenalty parameter. Therefore, only single objective optimization techniques are available inside the Approximation Loop and Pointer-2 strategies. These techniques are not designed to optimize several objectives simultaneously and require that a single objective function be formed from all selected objective parameters.

The value of the objective function is calculated as a sum of all objective components with corresponding weight and scale factors:

$(O b j e c t i v e = S u m ({O B J}_{i} \times W_{i} / S_{i}),$

where $O B J i$ is the contribution to the objective function of the $i - t h$ objective component (parameter), $W i$ is the corresponding weight factor, and $S i$ is the corresponding scale factor.

If the direction of a specific objective parameter is “minimize,” its contribution to the objective function equals the parameter value itself:

${O B J}_{i} = {P a r m}_{i}$

If the direction of a specific objective parameter is “maximize,” its contribution to the objective function equals the parameter value multiplied by –1 to reverse the direction, because the Optimization component always minimizes the objective function:

${O B J}_{i} = {P a r m}_{i} = - 1.0 \times {P a r m}_{i}$

If the direction of a specific objective parameter is “target,” its contribution to the objective function is calculated as follows:

${O B J}_{i} = {S q r t [(P a r m}_{i} {- T a r g e t}_{i})^{2} + 0.04] - 0.2$

The second component of the ObjectiveAndPenalty parameter is the penalty function, calculated as follows:

$P e n a l t y = P e n a l t y B a s e + P e n a l t y M u l t i p l i e r \times S u m {(V i o l a t i o n}_{i} \times W_{i} / S_{i}) \land P e n a l t y E x p o n e n t$

where $V i o l a t i o n i$ is the $i - t h$ constraint violation value, $W i$ is the corresponding weight factor, and $S i$ is the corresponding scale factor. The PenaltyBase option is set to zero if no constraints are violated. The default value is 0.0.

Double-click the Exploration component icon .
The Exploration Component Editor appears.
From the Exploration Component Editor, click the Objectives tab.
Right-click in the table to access various options for working with objectives.

For more information, see Setting Table Options.
Perform the following actions, which may vary based on your model design:
1. Determine which parameters you want to use as objectives by selecting the corresponding check boxes in the first column. Alternatively, you can click Check to add all the selected parameters (including array elements) as objectives. To clear all the parameters, click Uncheck.
  
  If no parameters are selected, you will be prompted to add all parameters as objectives.
  
  When a subflow parameter is selected as an objective, a corresponding parameter is created in the Exploration component.
2. Click the Direction column to specify whether or not you want to minimize the objective, maximize the objective, or set a target for the objective.
3. If you selected target in the Direction column, set the objective’s target in the corresponding column.
4. Set the objective’s Scale Factor and Weight Factor in the corresponding columns.
If desired, click Edit at the bottom of the editor to set attribute values for multiple objectives.

For more information, see Editing Attributes for Multiple Parameters.
If desired, map any of these objective attributes to parameters.

For more information, see Mapping Options and Attributes to Parameters.
If desired, click Disable strategy (execute a single run) to disable the Approximation Loop strategy when the Exploration component executes and instead execute the subflow once.
Click OK to save your changes and to close the Exploration Component Editor.