Open riga opened 4 months ago
I think the current choice of convention is to always skip the toinf
part and always keep the lower interval (e.g. dy_lep_pt0To50
(which is the only process with an interval starting at 0 as far as I see).
I have no strong opinion whether to keep the 0to
or the toinf
, but the 0to
part would be shorter and as far as I see there are rather few samples that start at a threshold of zero.
Keeping both parts would of course also be an option to be as verbose as possible. This would also clearly separate interval cuts from cuts that pick a singular value (e.g. 1j
, selecting events with exactly one jet)
In any case, I agree that we should decide on one convention and start enforcing it.
Datasets sometimes have closed intervals (
pt30to50
) or open ones (pt50
). The latter case usually meanspt ≥ 50
so that we could name itpt50toinf
.pro
It could be better to make the phase space of a dataset fully obvious. Especially since there might be cases where a cut is left-sided so that, say,
mgg50
would meanmgg ≤ 50
and thusmgg0to50
rather thanmgg50toinf
.con
Names are getting longer.
In any case, we have to find a consistent naming scheme. There are already some inconsistencies in the current master, especially in QCD datasets and processes, but also in DY, that need to be fixed once we defined a common scheme.