Hello everybody,

I have a question considering polyharmonic splines:

We use polyharmonic splines for approximation/interpolation of a 3D function with really good results. When doing an approximation, we are using a smooth factor such as one can do with thin plate spline.
It is working.
However, I have not seen any paper or report that concerns the smoothing fact.
Since thin plate splines are a special case of polyharmonic splines, it is not obvious that it is mathematically correct to do the same thing with all polyharmonic kernels... (even if it looks ok to me to do it)

If someone has some hints concerning the correctness of this, some explanations or some readings for me... It would be nice.

Thank you.