Skip to content

Making electron LH tool ready for rel 22 e/y ambiguity

Philip Sommer requested to merge ATLAS-EGamma/athena:philip_electronLikelihoodAmbiguity into master

The electron ID cuts on the AmbiguityType. When we added a new type in master, we hard coded the new type to always pass in ElectronSelectorHelpers. In rel 22 we want to be able to cut it, so the hardcoded pass of AmbiguityType==5 has to go (that's what the 32=2**5 means in ElectronSelectorHelpers).

The issue is summarised in @christos comment in ElectronSelectorHelpers.cxx which I remove in this MR:

  // We want reject types
  //  ambiguousVertexEoverPBetterThanTrackEoverP=6,
  //  photon=7,
  //
  // What pass is 0b11111 (31) e.g we do not specify explicitly the bit
  // to cut on. Which of course now breaks.
  // The magic 31 comes from config files which will be updated
  // at some stage for new tunes.
  // And the code below pass for types 0,1,2,3,4 (the 5 bits sets)
  // 2^0 + 2^1  + 2^2 +2^3 +2^4 = 1+2+4+8+16 = 31
  //
  // For now solve ad-hoc and keep compatibility with existing config
  // tooling in GroupArea/PHYS/Derivation etc by adding 32

I've created new config files for AsgElectronLikelihoodTool which add the 32 (pass type-5) to the cut on the AmbiguityType. This MR:

  • removes the hardcoded pass on type-5 in ElectronSelectorHelpers
  • adds the pass on type-5 in the configs of AsgElectronLikelihoodTool

So, this MR looks much more dramatic than it is! If I did everything correctly, nobody should see a change!

In case I made a mistake, I'm mentioning a few people, so everybody is aware:

Cheers, Philip and @christos

Merge request reports