dnorm/Lagrange 5 parameters does not come from books of "Grand Savants", just from my box.
Generally, the native "Gaussian ksmooth" is sufficient, useless for Carbon 12 α.
The J(x) fraction is from the conversion of either "continued fraction" or from
Padé normalised rational fraction. This algorithm is not new, it dates back to the times
of the first IBM MainFrames. Maple has maintained it and is accessible from Smath.
The most gorgeous piece of maths in there is the approximation of the smoothed data sets
to the rational fraction. The original algo is collaboration Jean/Robert
[Mathsoft Collaboratory ... 2003]

Enjoy, comment ... Jean

Smoothlet Splres Copy.sm (37 KiB) downloaded 71 time(s).

Smoothlet Supersmoothlet_Lagrange C12 Copy.sm (71 KiB) downloaded 69 time(s).