Package jdistlib.inference

Interface ParameterConstraint

public interface ParameterConstraint
A differentiable map from unconstrained coordinates to constrained values.

  • Method Summary

    Modifier and Type
    Method
    Description
    double
    constrain(double[] source, int sourceOffset, double[] target, int targetOffset)
    Constrains a slice and returns its log absolute Jacobian determinant.
    int
    constrainedDimension()
     
    String
    description()
     
    void
    pullback(double[] unconstrained, int unconstrainedOffset, double[] constrained, int constrainedOffset, double[] constrainedGradient, double[] unconstrainedGradient)
    Pulls a constrained gradient back and adds the log-Jacobian gradient.
    void
    unconstrain(double[] source, int sourceOffset, double[] target, int targetOffset)
    Maps constrained values back to unconstrained coordinates.
    int
    unconstrainedDimension()
     

  • Method Details

    • unconstrainedDimension

      int unconstrainedDimension()

    • constrainedDimension

      int constrainedDimension()

    • description

      String description()

    • constrain

      double constrain(double[] source, int sourceOffset, double[] target, int targetOffset)
      Constrains a slice and returns its log absolute Jacobian determinant.

    • unconstrain

      void unconstrain(double[] source, int sourceOffset, double[] target, int targetOffset)
      Maps constrained values back to unconstrained coordinates.

    • pullback

      void pullback(double[] unconstrained, int unconstrainedOffset, double[] constrained, int constrainedOffset, double[] constrainedGradient, double[] unconstrainedGradient)
      Pulls a constrained gradient back and adds the log-Jacobian gradient.