| Author |
 Topic  |
|
|
Docdoc
2 Posts |
 Posted - 12/09/2015 : 03:55:59 AM
|
Origin Ver. and Service Release (Select Help-->About Origin): 9.1.0 (64-bit) Sr3
Operating System: Windows 7
Hi,
I'm experiencing an issue with fits that are based on a fitting function that I have defined. In 9/10 cases the fits seem reasonable, but every now and then (1/10 cases) the fits are completely unreasonable. I have not been able to figure out why this happens, and why it happens only in 10% of the cases. I hope that someone here can help me.
The equation I want to fit is shown below. The values I want to obtain from the fit are "ro" and "s".
The fitting function I have defined in Origin looks like this:
The unreasonable fits usually look something like the fit shown below. As can also be seen the error for parameter s is really large. I have tried deleting values in the data table but the fits still look the same.
For comparison, most fits look something like this:
I would be very grateful if someone could help me! |
|
|
KenLefebvre
USA
66 Posts |
Posted - 12/09/2015 : 09:49:19 AM
|
Hey Daniel-
Off the top of my head I'm not sure why this is happening with such irregularity. Two questions come to mind- have you tried adjusting the Max Number of Iterations in Settings>Advanced>Fit Control and if so has this affected the frequency of these cases? And secondly, is there anything unusual about the data you're working with (e.g. size, format, disparate number of significant digits between cells)?
With thanks,
Ken Lefebvre
Technical Service Representative
OriginLab |
 |
|
|
Docdoc
2 Posts |
Posted - 12/09/2015 : 10:51:28 AM
|
Hi,
The iteration max is set to 400 and usually only 30-40 iterations are needed to reach the chi-square limit (set to 1E-9). So the fits always converge, according to the software...If I do one iteration at a time, the first few iterations don't result in any abnormalities, but they appear after 4-5 iterations.
There is nothing unusual with the data, as far as I can tell. I have even manually entered the numbers into a new worksheet and new workbook, but the problem remains. I have also tried deleting values, without any success.
The number of significant digits varies from 1 to 3 in the cells, but that is the same as in other datasets that don't result in strange fits. In terms of size there is no general difference between datasets that work and those that don't work.
Any other ideas on what I could try? |
|
|
| |
Topic |
|
|
|
Issue with user-defined non-linear fits