[Date Index] [Thread Index] [Author Index]

Splines for smoothing data

• To: mathgroup at yoda.physics.unc.edu (Mathgroup mailing list)
• Subject: Splines for smoothing data
• From: David Harrison <harrison at faraday.physics.utoronto.ca>
• Date: Tue, 29 Jun 1993 09:38:16 -0400

A common use of cubic splines in Physics is to smooth the data.  This
is, of course, very different from using them to 'connect the dots'
as in the NumericalMath`SplineFit` package with a Cubic option.

The sort of function I am seeking allows one to select any knots,
whether there happens to be a datapoint there or not, fit each segment
between the knots to a cubic polynomial in the independent variable,
and adjust the polynomials at the knot positions so they are continuous.
It is important that we are returned the sum-of-the-squares and the slope
at the knots since we will be doing a 'spline on spline' fit of the
slopes vs. the independent variable.

An example of such a program is the NAG routines E02BAF and E02BCF.

Does anybody have a Mathematica package that does this?  Failing that
a nice implementation of B-Splines would save me some coding.
--
David Harrison                             | "God doesn't play dice with the
Dept. of Physics, Univ. of Toronto         |   Universe!"  -- Einstein
Inet: harrison at faraday.physics.utoronto.ca |  "Stop telling God what to do."
Tel: 416-978-2977  Fax: 416-978-5848       |               -- Bohr