I’m hoping that someone with some experience with TFractionFitter can answer if the following can be done:
Can I fit multiple MC fractions (say three or more) where at least two of them have known relative ratios, but unknown total normalization? If so, how?
There are these “weight factor” histograms, but it is not clear to me exactly how they work.
I’m afraid that the present code does not allow to make such constraints. There is no particularly good reason for this: at the time of writing the code, the possibilities for making constraints like the ones you mention seemed sufficiently endless that I did not even attempt it.
It is not very hard to do what you want (but I admit not having tried this explicitly):
- make a parameter transformation such that the fractions of the relevant template sources are steered by a single fit parameter.
- then, in TFractionFitFCN() (or alternatively, in TFractionFitter::ComputeFCN()), make the reverse parameter transformation to make sure that in the likelihood computation proper “fractions” are used.
I’d be more than happy to put in any code allowing to generalize this, but I still think that is nontrivial.