Spectrum Viewer: Fitting Constraints #2842
Description
Current Behaviour
Currently the user can only set initial parameters for the fit
Desired Behaviour
Give the user the ability to limit a fitting parameter to a range or fix it to a specific value.
scipy.optimize. minimize has a bounds parameter that allows this (we are using Nelder-Mead which support bounds). The Bounds docs suggest that setting upper and lower to the same value fixes it.
Supporting this may also simplify the pre-fitting steps for Santisteban. Prefitting is running the fitting algorithm with only some of the parameters allowed to vary.
Undesired Behaviour
Importance of Feature
Key feature for 3.1
Additional Context
This is useful in various cases:
- In a noisy dataset bounds might be needed to prevent the optimiser ending up with a bad result
- For some experiments the user might know the value of one of the parameters but be interested in how another changes. E.g. the height of the edge at a specific wavelength may relate to the concentration of a material.