| Author |
 Topic  |
|
|
nkourkou
Greece
2 Posts |
 Posted - 01/14/2003 : 12:17:42 PM
|
Hello. I am trying to use the NLSF module with a user-defined function. I am following the tutorial, studying the FITEXMP3.OPJ file. Trying to apply this tutorial with a dataset of mine results in an unexpected failure.
The problem exists at "Action:Simulate". When I am defining my own equation (having 8 parameters and 5 independent variables) instead of getting theoretical values (to be compared with the original) I am getting a column full of dashes "-".
I am suspecting wrong initial values... Could it be that or it's that the parameters are too many?
(as a logical result, fitting produces error 28037)
Thank you in advance! |
|
|
easwar
USA
1964 Posts |
Posted - 01/14/2003 : 4:11:22 PM
|
Hi,
The fact that you are getting only dashes and no values indicates that there is something wrong in your function definition - in other words, your function is not generating values.
Also, note that currently the simulate feature is only supported for one independant variable. When there are more than one independant variable, and each is allowed to vary within a range, the resulting data becomes multidimensional and is not supported currently.
What you could do is the following: instead of using simulate, directly go to the fit page, assign some initial values to your parameters, and then click the "Chi-sqr" button. This will just generate the fit curve with your initial parameter values and will not do any iteration. You can then see if you are getting a good initial estimation, and you can also see at this point if your funciton definition is not generating values because of say a typo in your function definition etc. This will allow you to debug your function.
What version of Origin are you using? In version 7, there is support for defining user-defined functions using Origin C, and there is better debugging support.
Easwar
OriginLab. |
 |
|
|
nkourkou
Greece
2 Posts |
Posted - 01/15/2003 : 03:15:32 AM
|
Thank you for the instant reply... I am using Origin 7, my set of equations pass the "Compile before use" check and the setting below is "Equation".
Unfortunately the reduced Chi-sqr is 14.98...
Although I don't know what to do about the dashes problem, is it possible that the error comes from a wrong initialisation?
Thanks again...
|
|
|
|
easwar
USA
1964 Posts |
Posted - 01/15/2003 : 09:12:02 AM
|
Hi,
If you are getting a chi-sq value, then your function should be generating values. So it is not clear why you see dashes. Please send your Origin project and FDF file etc. to your local tech support representative, with a description of your problem.
Easwar
OriginLab.
Edited by - easwar on 01/15/2003 09:16:47 AM |
|
|
| |
Topic |
|
|
|
Nonlinear Least Squares Curve Fitter