Created on 2023-11-24 14:45:01 by Didier Vibert. % Done: 0

This concerns only the hybrid fitter and the derived lbfgsb fitter.
The individual fitter by definition is fitting each line-element individually (independently).
The svd and svdlc fitter are already fitting all the element jointly (the latter, svdlc, adding the joint fit of the continuum).
=> we should test the SVD fitter in line ratio mode, instead of individual fitter to fit jointly the 2 global ampitudes of emission and absorption. It will require more CPU, but I guess not too much...

Currently the hybrid fitter is launching the svd (or lbfgsb) fitter on lists of line elements sharing pixels. But the grouping of the line elements is restricted to group line elements of the same type, leading to fit independently emission and absorption elements that overlaps (See CLineModelElementList::getOverlappingElements).

Note: the hybrid fitter has also an option "improveBalmerfit" that corrects a-posteriori the superposed emission and absorption lines (currently only the Balmer lines: Halpha,beta,gamma,delta), if the emission absorption width are significantly different.

If we activate the joint fit of emission and absorption, then it should be exclusive of the "improveBalmer" option.
=> 2 parameters, or one parameter with 3 options.

@vlebrun, are there other lines that have emission and absorption which could be added to the "improveBalmer" option ?
Looking at the linecatalog, I see:

• the Paschen series P5_beta,6_gamma,7_delta,8_epsilon,9,10,11 (Palpha 18756.1 is only listed in emission in the linecatalog)
• the other Balmer lines: Hepsilon,8,9,10,11
• doublet MGII 2804.29 & 2797.11
• doublet CIV bizarre les longueurs d'ondes emission et absorption sont légèrement différentes: Abs = 1550.79 & 1548.19 vs Em = 1551.19 & 1548.61 ???
• LyAlpha,beta,gamma
• OVI 1037 & OVI 1031