Opened 18 years ago
Closed 18 years ago
#65 closed defect (fixed)
fitter fixed parameters not working
| Reported by: | Owned by: | Malte Marquarding | |
|---|---|---|---|
| Priority: | normal | Milestone: | ASAP 2.1 |
| Component: | General | Version: | 2.0 |
| Severity: | major | Keywords: | fitting |
| Cc: | awalsh@… |
Description
The other thing is I have noticed some strange behaviour when setting initial guesses. For something like "peakfixed=0", I find the behaviour changes depending on the number of gaussian components I am fitting. If I have one gaussian component, it all works fine if all the numbers are 0, as it should happen. However, when I have two components, I need to set the first component to "peakfixed=1" and the second to "peakfixed=0" in order to allow both peak heights to be flexible. ie. if the first "peakfixed=0", the the height for that component is fixed!
If I go to 3 gaussians, it gets even more complicated. To allow all components to be flexible I need the following:
component 0: peakfixed=0 component 1: peakfixed=1 component 2: peakfixed=0
I have track it down. The fixed argument needs to be inverted to for the underlying fitter. This inversion wasn't done when the fixed parameters are retrieved from the fitter, hence the even/odd behaviour.
I also fixed the 'fhwm' dislexia attack.