L2 inside-out
Compare changes
A new functionality of HLTMuon, L2 inside-out, is developed to recover inefficiency against close-by muon events at L2 level.
This can be turned on when a b-phys trigger chain contains "l2io" in it's names.
It doesn't affect the other chains, nor add any variables into xAOD HLTMuon objects.
It creates new dedicated containers only when running with AODFull.
It creates new dedicated containers of xAOD HLTMuon objects. When no "l2io" chain in menu, empty containers are created.
This MR includes adding three test L2 inside-out chains in LS2_v1.py; HLT_mu4_l2io_L1MU4, HLT_2mu14_l2io_L12MU10, and HLT_2mu6_l2io_L12MU6
The performance of the L2 inside-out is discussed in ATR-21003. Efficiency recovering was confirmed in the JIRA ticket. The CPU cost will be estimated in the same ticket after this MR. We can tune some criteria for the L2 inside-out to decrease CPU costs, or increase efficiency. Optimization studies can be performed based on the estimation of CPU costs. Then actual L2 inside-out chains for Run3 will be also discussed.
tagging @nakahama , @cdiez , @sshaw, @shhayash , @ktaniguc and @yuji