| Author |
 Topic  |
|
|
sperling
14 Posts |
 Posted - 07/06/2015 : 11:35:51 AM
|
I have a fitting problem that appears to be difficult to implement in Origin.
I have a set of absorbance spectra Ai(x) at various concentrations ci and discrete xj values. They unfortunately have baseline problems that prevent a straight-forward implementation of Beer's law, i.e. A(x)=a(x)cl, where absorbivity is a(x) and l is the path length. (Absorptivity is what I want out of the fit.)
While the usual spectroscopic methods for baseline subtraction work reasonably well, they have their own problems. I've found the baselines are well-described by parabolic functions, so each spectrum can be represented as
Aij = ajcil + fi + gixj + hixj^2
I've generated synthetic spectra that obey this relationship, and Excel solver lands on the correct values. Implementing this in NLFit in Origin 9.1, however, is problematic. I've tried various ways to use global fitting, but I have yet to figure out a way to couple the parameters to get (f,g,h) for each i and aj for each xj.
Is this even possible? Am I missing some trick to make this happen? |
|
|
jasonzhao
China
262 Posts |
Posted - 07/07/2015 : 01:47:20 AM
|
Hello,
Would you please introduced the detail to us, how do you implementing this in NLFit in Origin 9.1, and what's the problem.
You can illustrate your steps with screenshot, and send the data, OPJ, fdf to tech@originlab.com
You may need to mention this Forum post link in your email.
Best regards!
Jason
OriginLab Technical Service
|
 |
|
| |
Topic |
|
|
|
Difficult coupling problem for NLFIT