Skip to content

Options for removing tracks associated to muons in fullscan HLT PF rebuilding

Jon Burr requested to merge jburr/athena:HLTMuonPFTagging into master

This adds two ways to remove tracks associated to muons in PFO rebuilding. These can be conditionally enabled by using a new flag added to the trigger configuration Trigger.FSHad.PFOMuonRemoval which can take one of three options: "None", "Calo" and "Iso". Right now the default is "None" which should give the same results as the current master.

  • "None" is the current behaviour, in which no attempt is made to remove tracks that could be muons.
  • "Calo" uses the muon calo-tagging tools MuonCombinedCaloTag.pdf
  • "Iso" uses isolation-related selections, based on selections from METAssociation MuonTag.pdf

In order to use this in the HLT, we have to make sure that we only run the track extension once per track. For this, I had to change a few tool interfaces and add a new algorithm

  • The new algorithm PreselCaloExtensionBuilderAlg is responsible for building a calo extension for a view container. The current algorithm cannot work for a view container (which does not track indices) and also applies its own selections.
  • I had to modify the interfaces for the muon ID tools to accept an optional extension cache and cell container. This should not affect any exist usage (as these arguments default to nullptr) but tagging @jojungge and @wleight for approval
  • I also had to make a similar modification to the internals of the xAOD::CaloIsolationTool
  • Most of the new logic is implemented into new algorithms in the eflowRec package. The "Calo" option requires passing a cell container into this function. Right now this is hardcoded for the jet slice code but @jbossios and @valentem may have an idea of how to best avoid this.

Also tagging @khoo and @bcarlson for info.

Merge request reports