🚀 go-pugleaf

I have some old code from another person that used old IMSL routines to do
some cubic spline fitting. One IMSL routine was a bicubic spline 2D calculator
and another found the least squares approximation by cubic splines with variable
knots. 

In the newer IMSL these particular cubic splines have been replaced by B-spline 
routines. The 2D bicubic spline routine is replaced by a 2D tensor-product
spline interpolant that asks for where the knots are (the old routine did not
have any knots!). The lst sq cubic spliner is replaced by a B-spline least
sqr. These B-spline routines ask for a spline order and the number of
coefficients in the spline.

If I wanted to keep the cubic spline-like behavior, would the number of
coefficients = 4 and the spline order = 2? Our library has all the copies of
de Boor's book A Practical Guide to Splines checked out, so I'm asking the
net.  

Please respond via email:  strobel@dirac.phys.washington.edu

Thanks,
Nick
strobel@dirac.phys.washington.edu