Skip to content

Draft: Fixes for SDO creation during MC Overlay jobs

John Derek Chapman requested to merge jchapman/athena:ATLASSIM-7247_main into main

In ATLASSIM-7247 it was reported that on 10th July 2024 SDO creation began to fail in MC Overlay jobs run in the main nightlies - it works in 24.0 though. I suspect that this is related to some aspect of the barcode migration, but cannot find the exact cause. SDO creation still works in pile-up digitization, so there must be another cause.

Initially it was noted that CopyMcEventCollection produced an McEventCollection with different event numbers than SimpleMergeMcEventCollTool. This has been partially rectified, but the behaviour of CopyMcEventCollection for "separator" GenEvents still needs to be cleaned up.

Next it was noted that the Bkg_TruthEvent McEventCollection was being picked by HepMcParticleLink as the McEventCollection to which all links should point. In the case of MC Overlay jobs (without pile-up truth) then the HepMcParticleLinks in the input SimHits should point at Sig_TruthEvent, while the SDOs should point at TruthEvent. First Sig_TruthEvent was added to the list of possible keys. This had the effect of making all links pointat Sig_TruthEvent (even the redirected ones). Tweaking the static HepMcParticleLink::getRedirectedLink method to re-roll the SG::DataProxy finding allowed the redirected links for hard-scatter GenParticles to point at TruthEvent, although the links to delta-rays still point at the Sig_TruthEvent.

Issues to resolve:

  • correctly re-direct delta-ray HepMcParticleLinks
  • correctly handle HepMcParticleLinks to pile-up GenEvents
  • further update the behaviour of CopyMcEventCollection.

Merge request reports