| Author |
 Topic  |
|
|
jmb22
USA
Posts |
 Posted - 01/20/2005 : 6:40:50 PM
|
Origin Version (Select Help-->About Origin): 7.0
Operating System: WinXp
Hi,
How does the fitting tool works?
I've got experimental data (about 10 points) that I want to use to extract 2 parameters using Analysis > Non-linear curve Fit > Advanced Fitting tool.
To be more precise, I plotted frequency F (in 1E9Hz up to 24E9 range) versus field B: F vs. B. The data must fit with the following equation:
y = P_1*sqrt(x*(x+P_2))
where y is the frequency, x is the field and P1 and P2 are the fitting parameters.
I obtained then:
P_1=10.46 x 10^12
P_2=12597
Here is the problem:
If I replace y (Frequency) by y=(Frequency)^2, then the Equation that I want to use to fit the data F^2 vs. B is:
y = (P_3)*(x*(x+P_4))
I obtained then the following fitting parameters values:
P_3 = 3330033
P_4= 11743
Those values do not match the values from the previous fitting equation. I was expecting to get at least P_2 = P_4
Anyway, why don't I find consistent fitting parameters using the same equation written differently?
Could you advice the best way to do that kind of non-linear curve fit?
Tahnx for the help.
Best
Jean-Marc
|
|
|
easwar
USA
1965 Posts |
Posted - 01/21/2005 : 10:43:06 AM
|
Hi Jean-Marc,
If I understand this correctly, you are first fitting with one equation to your y data, and then you create a new dataset which is (your old y data)^2 and then you fit it with the second equation. If this is the case, the fitter is not working with the same data/parameter space in both cases and so the fit results could be different. And the difference is probably due to errors in the data, as in the fit in either case is not a "perfect" fit to the data. You did not quote what kind of parameter error values you get in each case. The values you get for the parameters in either case could be within the parameter errors.
I suggest you send your data to tech support and refer to this forum post, so that someone can look at the data and fit with your equations etc. You could also send your OPJ and your FDF files.
Easwar
OriginLab
|
 |
|
|
jmb22
USA
Posts |
Posted - 01/21/2005 : 11:53:33 AM
|
Hi Easwar,
Thanx for the suggestion.
By the way, I didn't have time to send you more data regarding a previous post (can't remember the title). It was about a 2 level folder creation depdending on file's name.
If needed, I can send you something by tomorrow.
Best
jean-Marc |
|
|
| |
Topic |
|
|
|
fitting tool : intrinsic properties ???