Skip to content

AnalysisTop 21.2: Add fJVT working points and scale-factors

Jonathan Jamieson requested to merge jojamies/athena:21.2-AT_ImplementfJVT into 21.2

JIRA ticket: ANALYSISTO-973

Adds 3 config options for forward (eta>2.5) jet-vertex-tagging comparable to existing options for JVT:

  1. ForwardJVTWP - None (default), Tight, Medium Sets related fJVT working point and will automatically remove forward jets from the event if they fail the WP cut (>0.4 for Tight, >0.5 for Medium). Also builds nominal+systematic event weights from the fJVT scale-factors of each jet into 3 new branches: weight_forwardjvt, weight_forwardjvt_UP, weight_forwardjvt_DOWN If None options is used, neither the fJVT selection tool, nor the fJVT efficiency tool will be initialised, all forward jets in event will be kept, and weight_forwardjvt branches won't be written out The Medium WP is recommended by the Jet/MET group however if this is also used within the METmaker then the MET working-point must be changed away from its recommended value of Tight to Tenacious. A warning is given in this case but as this is easily missed and essentially changes another parameter invisibly my preference is to recommend the Tight WP for fJVT instead.

  2. ForwardJVTinMETCalculation - False (default), True   * Replaces existing FwdJetAndMET config option *

    Enables the removal of fJVT_fail jets from the final calculation in the MET maker to improve MET resolution. Enabled on its own (i.e. with ForwardJVTWP = None) this option uses the Tight working point, and does NOT run the fJVT efficiency tool or build the event weight branches. This is to ensure users do not incorrectly weight events which do not actually cut on fJVT. The effect ForwardJVTinMETCalculation = True will have on the final event numbers via any MET cuts is not simple and would need to be calculated on a case-by-case basis which is not usually something which is considered.

  3. SaveFailForwardJVTJets - False (default), True Produces a separate jet collection to save forward jets which fail the fJVT cut, mostly for debugging purposes.

Two additional considerations:

If the fJVTWP option is used, the fJVT decision will be used to change the priority of failing jets within the overlap removal, as is done for JVT. (With a lower priority, other physics objects can remove fJVT_fail jets, but not the other way around)

Lastly the fJVT decision for PFlow jets must be calculated at derivation level thus a check is put in place to ensure users requesting fJVT for PFlow jets are using a viable derivation (>=21.2.97.0)

Jet and MET comparison plots for both EMTopo and PFlow jets: EMTopo_60K_Events_2016_MC   PFlow_6K_Events_2018_MC_NightlyDerivation

Yellow: ForwardJVTWP = None, ForwardJVTinMETCalculation = False

Black dotted: ForwardJVTWP = Tight, ForwardJVTinMETCalculation = True

Edited by Jonathan Jamieson

Merge request reports