VVE Functionality - calcluation of coeffs after editing VVE graph

I have wasted a lot of time trying to get a smooth VVE for BOTH the 0,0 Cam angles, AND 20,20 Cam angles.
I am becoming convinced that what some users are saying: i.e. that the way the coefficients are solved for is not exactly correct, is right.

I'm going to have to try to program the VVE myself in R or Python.

Can anyone tell me what, exactly, the equations/functions are used to generate the VVE (as a function of the MAP, PR, MAP^2, and the Boundaries)?
Also, are the any transformations applied to the input variables (MAP, PR, MAP^2, and the Boundaries) before they feed into the function/equations?

Even better, does anyone know what type of surface-fitting algorithm that GE actually used to create the VVE? Was the VVE created using a NURBS surface fitting model?

Or, even even better, does anyone have any Matlab code that they used to do solve for their own VVE coefficients (i.e. what the Bluecat software does)?


Thanks very much.