Add latest and streamlined implementation for OS taggers
this merge request propagates the latest versions of the OS taggers (currently living in dev-flavtag-ignorexgb
and dev-flavtag-master
) to 2018-patches
in order to put them into production.
The last thing that @kheinick and I would like to discuss with you is the naming scheme for the branches in the tuples. At the moment, there are two "nominal" versions that can be used.
If you use this:
applyFTTuning(btagtool, tuning_version="Summer2017Optimisation_Run1")
then you get the nominal Run 1-optimised taggers in your tuple (where only OSe has an update):
"OSCharm"
"OSElectron"
"OSElectronLatest"
"OSKaon"
"OSMuon"
"OSVtxCh"
"SSKaon"
"SSPion"
"SSProton"
If you use this:
applyFTTuning(btagtool, tuning_version="Summer2017Optimisation_v4_Run2")
then you get the nominal Run2-optimised taggers together with the "old" taggers:
"OSCharm"
"OSElectron"
"OSElectronLatest"
"OSKaon"
"OSKaonLatest"
"OSMuon"
"OSMuonLatest"
"OSVtxCh"
"SSKaon"
"SSKaonLatest"
"SSPion"
"SSProton"
Basically, the general rule for the analyst is "choose the Latest whenever you see it in the tuple, according to the period of data taking (Run1 or Run2)".
We would like to know if you consider this naming as satisfactory, or if you think that this can be improved.
Once we take a decision, we will remove the WIP
and merge this branch.
This merge request will go in parallel with !1310.
In the meanwhile, the dev-flavtag-OSLatest
branch has been merged to dev-flavtag-ignorexgb
and dev-flavtag-master
, so you can follow these instructions:
https://gitlab.cern.ch/lhcb-ft/TupleProduction/tree/dev-b2cc-dimuon-newframework
as they are to produce tuples with the "temporary" recipe.