Skip to content

Add proto particle making to Hlt2 reconstruction

Sascha Stahl requested to merge sstahl_adapt_protos into master

The MR add proto particle making to Moore.

The functionality is implemented in the following files

  • protoparticles.py includes functions to create charged and neutral protoparticles from reconstruction output.
  • hlt2_global_reco.py sets up the whole reconstruction data flow. Its output can be used in particle making and selections.

Additionally, the Hlt2 Muon ID is added as it was missing before. It had to be disabled again, as it is not thread-safe.

The proto particle making is also added to hlt2_reco_baseline. It should not have an effect on existing tests besides adding new counters. A significant drop in throughput is expected.

Two tests are added:

  • hlt2_particles_baseline runs an options file which uses the proto particles from the reconstruction to make pions and muons and subsequently create Ks and JPsi. It is only checked that the job configures and runs. No reference testing.
  • hlt2_protoparticles_baseline mirrors the test brunel-upgrade-test-protos.qmt and tests the content of every neutral and charged proto particle. The Brunel test can then be removed.

Next steps for the next MRs:

  • Make it configurable to either use reconstruction from file or run reconstruction.
  • Make proto particle sequence functional.
  • Make proto particle sequence faster.
  • Create functor cache for running HLT2 reco. ProtoParticle making added a LoKi functor, therefore I removed the file disabling python functors when running HLT2 reco.
  • Change input file of brunelesque tests to be the same as in Brunel.
Edited by Sascha Stahl

Merge request reports