From: Prof.Dr. M.D. Lechner <lechner@rz.Uni-Osnabrueck.DE>
  To  : RASMB@BBRI.Harvard.edu
  Date: Mon, 10 Feb 1997 12:44:12 +0100

ORIGIN & Splines

Dear RASMBers,
H. Coelfen asks in his mail from Jan 31, 1997 for ORIGIN
applications applying spline functions. I guess he means
regression splines with virtual points and the condition
that the sum of the square root distances of the experimental
points from the spline function passing this virtual points
is a minimum.

I'm afraid that more sophisticated problems such as regression
splines does not work properly with ORIGIN. ORIGIN is a very
nice graphics program with access of experimental data. It
would be more convenient to use for sophisticated problems
a common programming language, e.g. Fortran, C++, Visual Basic
and then introducing the results into ORIGIN.

Back to regression splines for smoothing of data! In my opinion
this is a convenient method and we have developed a lot of
programs for evaluation of AUC measurements with smoothing of 
experimental data with regression splines. But you have to be 
carefully to arrange the number and place of the virtual points.
For small curvatures it may be convenient to use equidistant
virtual points whereas for big curvatures it would be better
to use curvature dependent distances of the virtual points.
In some complicated cases you may use a combination of 
equidistant and curvature dependent distances of the virtual 
points.

It would be very nice if anybody can give more information
about this problem.

Yours M.D. Lechner, Physical Chemistry, University of Osnabrueck,
      Barbarastr. 7, 49069 Osnabrueck, Germany