Like many astronomers, my numerical analysis
tool box is pretty much defined by the topics in
Press et al. "Numerical Recipes". Recently I had
some Monte Carlo data that I wanted to fit with
a spline and couldn't find that in NR. I.e., I
didn't want to pass the curve through the (noisy)
data points, I wanted a least squares fit. This
led me to the realm of curve and surface fitting,
and such dense things as "The NURBS Book". Since
I like to have the tools I'm using in J, rather
than call a routine in some other language (R?)
I was faced with a lot of learning & translation.
I finally found a source for the limited problem
I wanted to solve, and now have working routine.
(See my J page:
http://www.astro.umd.edu/~jph/ls_spline.ijs
This code is just translated from pseudocode and
is loopy and not J-like -- that's still to do.)
So I was just wondering if any in the wider J
community has produced code or experimented with
such things as B-splines, NURBS, etc. I realize
this is a huge area because of computer graphics
applications, but it would be good to have some
of the basic stuff implemented in J. (I did find
the script for the Levenberg-Marquardt Algorithm
on the wiki -- I've used that in IDL and it's
great to find a J version.)
Excuse the long post, but I didn't want to
reinvent something that was already out there.
Patrick
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
