Renormalisation Systematic #74
Conversation
|
I was originally going to have a check that this class of systematic was never applied to a PDF whose NormFittingStatus wasn't FALSE. However, it's a bit tricky to do this because the BinnedEDManager only knows the NormFittingStatus and not the systematics, and the SystematicManager knows about the systematic types, but the not the NormFittingStatus of the PDFs they apply to. Both sides just see a string of a group name From outside them, (eg from the BinnedNLLH class), I can't access all that information without adding to the code in inelegant ways, which I am hesitant to do I think as nothing bad would happen if you tried to apply this to a non-FALSE PDF (it would just have no effect), it's ok for this to go ahead and be reviewed. But I am open to suggestions. And of course there's no rush on this at all, I am not certain any one will ever use it now, but given I've coded it up we may as well merge it at some point |
This adds a new systematic class that just renormalises PDFs (so would be used on PDFs with norm fitting status == false)
It is equivalent to a shape systematic where the shape function would just return a scaling factor (scaling up and down not left to right)
There's some basic unit tests, but I think we want to think about this a bit before merging. We should probably at least have a check that any PDF it's used on does have norm fitting status == false, else the behaviour will likely not be what the user intends