Configuring the Importance Sampling Technique

You can configure the Importance Sampling Technique options.

  1. Double-click the Six Sigma component icon .

    The Six Sigma Component Editor appears.

  2. From the General tab's Analysis Type list, select Reliability Tehnique, and from the Technique list, select Importance Sampling.

  3. In the Technique Options area, select one of the following:

    Option Description
    Simple Importance Sampling Generates a set of designs around the most probable point to estimate the probability of failure.
    Adaptive Importance Sampling Samples in the failure region to compute the probability of failure.

  4. Enter or select the following:

    Option Description
    Correct FORM Estimate (Simple Importance Sampling only) This parameter selects the value to be used to populate the Six Sigma Results parameter. The probability of failure is calculated by Simple Importance Sampling if this option is not selected, and is calculated by the FORM correction method if this option is selected.
    Maximum AIS Iterations (Adaptive Importance Sampling only) This parameter represents the total number of Adaptive Importance Sampling iterations to be performed. Adaptive Importance Sampling approximates the failure region by a second order polynomial using principle curvatures and samples only in the failure region; the curvature values are updated iteratively to ensure that the failure region is adequately covered. The Maximum AIS Iterations value indicates the maximum number of such curvature updates to be performed in case AIS does not converge.
    Number of Sample Points This parameter represents the total number of sample points to be generated at the MPP after the MPP search for each response. The default is 50.
    MPP Maximum Iterations This parameter represents the maximum number of iterations that are run to search for the Most Probable Point (MPP) for each response with limits defined. The MPP is the most likely failure point for a given response.
    Finite Difference Step Size This parameter sets the relative finite difference step size to be used by the reliability method when calculating gradients. The default value is 0.01 (1%).
    Minimum Finite Difference Step Size This parameter represents the minimum absolute finite difference step size to be used by the reliability analysis method when calculating gradients. The default value is 0.001 (0.1%).
    MPP Absolute Convergence This parameter represents the maximum absolute change in the value of the distance from a constraint surface between two consecutive iterations of the MPP search. For example, if you wanted the search process to stop if the change in the distance from a constraint went below 0.01, you would enter 0.01. The default is 0.0001.
    MPP Relative Convergence This parameter represents the maximum relative change in the value of the distance from a constraint surface during the MPP search. The default is 0.001.
    Use fixed random seed When this option is selected, the Importance Sampling method uses a fixed seed value for the random number generator. If you do not enter a value, the clock time is used as the seed value.

  5. Click OK to save your changes and to close the Six Sigma Component Editor.