Move all pp HLT jet reco configuration to be based on JetDef
Following on from !70788 (merged), this is a cleanup of the HLT jet reconstruction configuration.
Previously we did all configuration based on the jetRecoDict, extracted from the chainDict. Since we factored out the definition of all JetDefinitions, we can now use those configuration objects to configure all algorithms directly without the further need to depend on jetRecoDict. This is more in the declarative spirit of the JetRecConfig package -- determine all your dependencies, and then step through and configure the components needed to satisfy them.
-
25.04.2024: This runs the jet slice -- consistency checking needed. The MET slice seems to generate as well with minor fixes.
-
Fix the Unconventional Track Trigger chains to use the new syntax
-
-
26.04.2024: Full menu now generates, but config differences are seen
-
Fix config inconsistencies wrt clean r25 nightly -
Update code comments and any other documentation -
Clean up code and add helpers to interpret the JetDefinitionespecially for repeated operations
-
FYI a few people who might be interested: @pakontax, @miochoa, @cantel. I decided we didn't really need more helpers but maybe will make notes of a couple of changes in how we get some information:
- If you want to know what the
trkoptis inBlahRecoCfg, checkjetDef.context - If you want to check generic information about what the jet collection is made from, string-matching
jetDef.inputdef.labelis quite flexible. - If you need to get access to the track collection names, grab them from
jetDef._contextDic, which is equivalent toflags.Jet.Context.[trkopt]. - To get the trigger jet calibration string, use
jetCalibFromJetDef(jetDef) - The default jet calibration getter has been changed to use the args
getJetCalibDefaultString(recoAlg, constitType, trkopt)for easier compatibility both withjetRecoDictand `JetDefinition. The way I call it with the latter is not super elegant, but it's also not too offensive.
@fpastore I did not attempt to use getStep yet -- this needs a big reorganisation in JetChainConfiguration for most cases (some other cases are trivial and I would just do them all together). I have some idea as to how it could work, but it's better to go progressively.
Merge request reports
Activity
FYI @delsart, a couple more tweaks in
JetDefinitionhere -- hashing ofJetInputExternaland translation ofxAODType.Jetas a constituent type.-
Full menu diff (intended change):
Component jetrecalg_HLT_AntiKt10EMTopoRCJets_subjesIS.builder may differ JetInputType = 12 vs 100 << but all are suppressed by renaming/known differences/... Component jetrecalg_HLT_AntiKt10EMTopoRCJets_subjesIS_ftf.builder may differ JetInputType = 12 vs 100 <<Edited by Teng Jian Khoo -
Added one more change affecting config:
Component TrigJetHypoAlg_a10_tc_em_nojcalib_calopresel may differ Jets = HLT_AntiKt10EMTopoJets_nojcalib_pt10 vs HLT_AntiKt10EMTopoJets_nojcalib << but all are suppressed by renaming/known differences/... Component TrigJetHypoAlg_a10_tc_lcw_subjes may differ Jets = HLT_AntiKt10LCTopoJets_subjes_pt10 vs HLT_AntiKt10LCTopoJets_subjes << but all are suppressed by renaming/known differences/... Component TrigJetHypoAlg_a4_pf_em_subjesgscIS_ftf may differ Jets = HLT_AntiKt4EMPFlowJets_subjesgscIS_ftf_pt10 vs HLT_AntiKt4EMPFlowJets_subjesgscIS_ftf << but all are suppressed by renaming/known differences/... Component TrigJetHypoAlg_a4_pf_em_subresjesgscIS_ftf may differ Jets = HLT_AntiKt4EMPFlowJets_subresjesgscIS_ftf_pt10 vs HLT_AntiKt4EMPFlowJets_subresjesgscIS_ftf << but all are suppressed by renaming/known differences/... Component TrigJetHypoAlg_a4_tc_em_subjesIS may differ Jets = HLT_AntiKt4EMTopoJets_subjesIS_pt10 vs HLT_AntiKt4EMTopoJets_subjesIS << but all are suppressed by renaming/known differences/... Component TrigJetHypoAlg_a4_tc_em_subjesIS_calopresel may differ Jets = HLT_AntiKt4EMTopoJets_subjesIS_pt10 vs HLT_AntiKt4EMTopoJets_subjesIS << but all are suppressed by renaming/known differences/... Component TrigJetHypoAlg_a4_tc_em_subresjesgscIS_ftf may differ Jets = HLT_AntiKt4EMTopoJets_subresjesgscIS_ftf_pt10 vs HLT_AntiKt4EMTopoJets_subresjesgscIS_ftf << but all are suppressed by renaming/known differences/...Somewhere in the CA migration we stopped passing the filtered jet collections produced by
JetViewAlgsto the hypo. No possible effect on the output decisions, just a sub-ms optimisation that went missing (it used to save significantly more). Re-adding this rather than taking out the JetView algs because technically it could save something of O(0.01%) of the CPU cost by running something that costs 10x less. But we could simplify config and save a small amount of memory by taking them out for quite a negligible cost.Maybe to look at cost data after this is merged to see if the impact of the correction is measurable.
