I previously posted a problem of mine regarding nonlinear regression with
the levenberg-marquardt method described in the numerical recipes in c book.
I was informed that this was a sloppy implementation of the method and not
very robust for nonlinear regression. I received good feedback from the
newsgroup especially from Petr Kuzmik and David Fournier from Otter
Research.
Petr ran some of my data through his code and returned estimated parameters
for me. Dave took the time out to address my problem and build a model for
me using his software, AD Model Builder which can be found here:
http://otter-rsch.com/admodel.htm. I highly recommend this software to
anyone who needs a FAST and ACCURATE model for nonlinear regression (I'm
sure it does other things too). I compared his results with Jandel's Table
Curve 2-D and results matchup to a "T."
Thanks to all that helped,
Robert
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
