home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math.symbolic
- Path: sparky!uunet!mcsun!sun4nl!alchemy!prisma!freek
- From: freek@cv.ruu.nl (Freek Beekman)
- Subject: Non-linear fitting with Mathematica.
- Sender: usenet@cv.ruu.nl (Usenet Network)
- Message-ID: <1992Jul23.090624.18725@cv.ruu.nl>
- Date: Thu, 23 Jul 1992 09:06:24 GMT
- Distribution: sci.math.symbolic
- Nntp-Posting-Host: drogo.cv.ruu.nl
- Organization: University of Utrecht, 3D Computer Vision Research Group
- Keywords: curve fitting, least squares approximation. non-linear
- Lines: 33
-
- Hello,
-
- who knows how to do a non-linear fit with Mathematics. I want to fit a
- non-linear function(A produkt of error-functions, Gaussians and
- exponential functions with the parameters to be fitted also in the
- argument of the functions). I think i cant reduce the function to a
- linear form as is shown in the Mathematics book. I have nine parameters
- to fit, and about 2000 measured points. The fit has to be a least square
- fit. If it is possible I want to know how I can have speed up this fit
- routine, because I feel it can take a lot of time. The derivatives are
- large formulas. Therefore it could take a long time to evaluate them in
- the
- fitroutine, and therefore it would be better to take a fit routine that
- doesnt use derivatives during the fit.
-
- Thanks in advance for your reply,
-
- Freek.
-
- --
- - Freek J. Beekman -
-
- Department of Radiology and Nuclear Medicine/3D ComputerVision
- Utrecht University Hospital, Room E02.222
- Heidelberglaan 100
- 3584CX Utrecht
- The Netherlands
- Phone +31-30-506711/506053
- Fax +31-30-513399
- email freek@cv.ruu.nl
-
-
-
-