Option to symmetrise the data is not ideal, and should be replaced by subtracting FSE to the first NCP peak
Errors on the parameters, especially for functions like GC_C4, GC_C6 and GC_C4_C6 are often on the same magnitude of the parameter values, which invalidates the fit
Global fit on all detectors is slow
Positivity constraint is not being successfully implemented, since some fitted functions have negative values
Possible solutions and avenues to explore:
[ ] Implement FSE corrections before focusing data
[ ] Instead of convolving NCP to fit, deconvolve the data and fit NCP
[ ] Explore using Mantid algorithm for plotting the error profiles of the parameters, maybe an alternative to using MINOS?
[ ] Use Mantid Fit for global fit instead of using iminuit
[ ] Use analytical solutions for the positivity constraints where possible, resort to brute force numerical range constraint for GC_C4_C6
Current routine has a few shortcomings:
Possible solutions and avenues to explore: