Overriding the Distribution Criteria for a Single Statistical Setup

Follow this procedure to override the default distribution criteria for a single statistical setup:

  1. In the Setup Statistical Analysis dialog box, click the Variables tab. All of the variables that were selected for statistical analysis appear.
  1. Select or clear the Include check box for each variable to define the specific variables included in the statistical analysis setup.
  2. For each included variable, select Uniform, Gaussian, Lognormal, or User Defined in the Distribution column for the variable you want to override.

    If you changed the distribution type, the Override option is now selected. This indicates that the distribution type you selected will be used for this optimization analysis; the current distribution type selected for the variable in the nominal design is ignored in this statistical analysis.

    • Alternatively, you can select the Override option first, then select a different distribution type in the Distribution text box.
  3. Optionally, if you want to change the distribution criteria, click in the Distribution Criteria column for the variable you want to override. The Edit Distribution dialog box appears.
  1. If the distribution type is Gaussian:
    1. Enter the standard deviation in the Std. Dev text box.
    2. Enter the lower limit of the distribution in the Low Cutoff text box.
    3. Enter the upper limit of the distribution in the High Cutoff text box.

    Twin Builder will solve design variations using a Gaussian distribution within the low and high cutoff values.

  2. If the distribution type is Uniform, enter a tolerance value in the text box.

    Twin Builder solves design variations within the tolerance range of the starting value, using an even distribution.

  3. If the distribution type is Lognormal:
    1. Enter the cutoff probability in the Cutoff Probability text box.
    2. Enter the sigma value of the distribution in the Sigma text box and select a unit from the drop-down list.
    3. Enter the m value of the distribution in the M text box.
    4. Enter the theta value in the Theta text box and select a unit from the drop-down list.
  4. If the distribution type is User Defined:
    1. Enter the cutoff probability in the Cutoff Probability text box.
    2. Click Edit XY Data to open the Edit Datasets dialog box in which you can select an existing dataset, or create a new one.
  5. By default, all variables are set to sample using Latin Hypercube sampling. This sampling method provides for greater variability than random sampling by keeping track of chosen samples and guaranteeing that samples cannot be repeated. Revert to random sampling by clearing the check box in the Latin Hypercube column for any desired variable.
  6. Click OK.

    To revert to the default distribution settings, clear the Override option.

Related Topics 

Statistical Cutoffs