Skip to content

L1calo efex tau bdt reduce eta asymmetry

Since the heuristic algorithm produces a small seed with values in [0,3] and uses that to give a fine resolution for the TOB eta while the BDT algorithm sets this small seed value to 0, there is a slight asymmetry in eta for BDT TOBs with the way it's currently computed. For example, here's a plot of delta eta between tau jet and matched BDT TOB:Pasted_image_20230921165032

To mitigate this:

  1. Set the default seed value for BDT TOBs to 1 instead of 0. L1Topo algorithms should not be affected for a hard-coded seed value of 1. The only use of this seed value (that I could find) at L1Topo is through the iEtaTopo() method here: https://gitlab.cern.ch:8443/atlas/athena/-/blob/main/Trigger/TrigT1/L1Topo/L1TopoAlgorithms/Root/cTauMultiplicity.cxx#L189 and here: https://gitlab.cern.ch:8443/atlas/athena/-/blob/main/Trigger/TrigT1/L1Topo/L1TopoAlgorithms/Root/cTauMultiplicity.cxx#L224.

  2. The eta() method of the eFexTauRoI EDM will return a coarser value of eta for BDT TOBs than for heuristic TOBs. This value will be the eta of the center point of the 3x3 rectangle.

  3. Also added enums for Heuristic and BDT algorithms for convenience and clarity in the tau EDM object.

Merge request reports