| Author |
 Topic  |
|
|
Bobby64
Spain
5 Posts |
 Posted - 09/10/2012 : 04:09:12 AM
|
Origin Ver. and Service Release: OriginPro 8.5
Operating System: Windows XP
Hi,
I would need some help about the non linear fitting procedure. I guess my problem isn't so complicated, but right now I just do not see how to solve it...
Anyway, here it is:
The next picture is an example of data set
Usually I perform a non linear fitting on this kind of data set, using the following equation (1): I=Imax*(1-exp(-(D+DE)/D0)) (where I is the dependant variable, D, the independent variable and Imax, DE and D0 are the fitted parameters). In addition, I always use a weighting of the data by the inverse of the squared intensities (1/I^2) (in Origin: arbitrary dataset, column B).
As an example, here is the result corresponding to the data set I previously detailed:
Until this step I do not have any problems. However, I recently read that it was possible to get a similar result by converting the previous function to (2): -ln(1-(I/Imax))=(D+DE)/D0. According to the authors, by plotting -ln(1-(I/Imax)) versus the dose (D), the data are converted into a straight line and the parameters result from simple straight line fitting.
So, I tried to see if I could get similar results with function (2) and I entered it in the fitting function builder, but I do not find the same results between functions 1 and 2, and in many cases I have an error message...
Could someone give me some advice about the way to proceed ?
Many thanks in advance,
Cheers,
Bobby64 |
|
|
Kathy_Wang
China
159 Posts |
Posted - 09/14/2012 : 04:01:01 AM
|
Hi,
-ln(1-(I/Imax))=(D+DE)/D0 is an implicit function, and implicit fitting is not yet supported in Origin 8.5 (but it will be supported in our next version 9.0).
With Origin 8.6 and earlier versions, it's only supported to fit explicit functions, i.e. the fitting function shall be in form of Y=f(X).
Kathy
Originlab |
 |
|
| |
Topic |
|
|
|
NL fitting: linear conversion procedure