I have noted recently that sometimes the amplitude of the lorentzians were underestimated, specially for narrow lorentzians. This is due to a sampling problem. Delta functions are computed as very narrow lorentzians which are then convoluted with the resolution (or another broder lorentzian like for example in the case of a rotational motion that is decoupled to a translation). I advise for now to set the center to 0 so this problem is not an issue. I am now testing a new version which will fix this problem and also give some new features such as confidence interval calculations.
Also available in: Atom