More Lev-Mar questions

[TheInfamousYano]

Hi,

I was doing a nonlinear Lev-Mar fit successfully with 4 parameters (thank you altenbach). Then I changed to 6 parameters, and now the original four have become static. In other words, when I run "VApprox.vi" on the sample data below, the program returns the initial guesses as the best fit coefficients for the first four parameters; whereas the new parameters (a & b) change.

Thanks

[Travis_M.]

Hello! 

I’m no expert on Lev-Mar, but I can say that this is one of the more complicated VIs to use in LabVIEW.  Have you checked out the documentation regarding the VIs you need to modify to get this one to work?  There are some good posts on these discussion forums from users working with Lev-Mar.  I’ll go ahead a put a few links below that might help out.

http://digital.ni.com/public.nsf/websearch/5BF55E4887A063B086256CAE0073A12E?OpenDocument

http://forums.ni.com/ni/board/message?board.id=170&message.id=51344&requireLogin=False

Sorry I don’t have anything more specific at this time, but I do hope this helps!

[medvall]

Hi Yano,

The best would be to try NLSSOL in TOMLAB /FITTING for LabVIEW. Then
you can also add bounds, linear/nonlinear constraints.

The package is currently available for beta testing (send email to me).

Best wishes, Marcus
Tomlab Optimization Inc.
http://tomlab.biz

[altenbach]

I have not had a chance to really analyze this, but it seems to me that the parameters are not very well balanced (for lack of a better term).

You can get good fits of all parameters if you e.g. multiply "a" by 10^-6. Try the modification to the model function shown in the image.

Message Edited by altenbach on 01-20-2006 03:44 PM