Kinematic-dependent prescales
The following discussion from !2314 (merged) should be addressed:
-
@graven started a discussion: (+5 comments) Re:
- Low pT -> 0-1.5 (prescale reduced to 0.002)
- Mid pT -> 1.5-3 (prescale set to 0.01)
- High pT -> 3-6 (prescale reset)
- Very High pT -> 6+
In general, IMO, it is better to 'nest' trigger lines in this type of scenario, i.e. define them as
Low: pT>0
,Mid: pT>1.5
,High: pT>3
andVery High: pT>6
. Now, that implies that the newly created overlap(s) is (are) triggered by multiple lines, and if the prescales are independent (which they are by default), this results in higher rates -- but the underlyingDeterministicPrescaler
s of these lines could be configured to use the sameSeedName
which then results in 'coherent' prescales, ie. the total rate for a set of lines with nested lower bounds will be the same as for a set of lines with non-overlapping lower and higher bounds -- and the nested lower bounds are much easier to use downstream, as one does not have to know the exact (relative) prescales of each subsequent line to properly normalize the spectrum.Note that this does require a bit of work, as right now the
SeedName
is always computed from the line name, and one would have to add a bit of code which checks if it is explicitly given, and if so, use that, and only if not use the line name as seed.