Skip to content

Draft: TOS match on track-by-track basis in MCTupleToolTOS

Ross John Hunter requested to merge rjhunter-TOS-algo-v2 into master

FYI @rmatev @sstahl @mvesteri

As described in the description of MooreAnalysis!33 (merged), MCTupleToolTOS (which matches the trigger candidate to the true MC signal candidate in the event to give a TOS-like efficiency) has a matching algorithm that will break down when it has to handle 3 or 4 track candidates. This is expanded upon in the associated vCHEP paper. Therefore for HltEfficiencyChecker to give TOS-like efficiencies for any HLT2 line, this envisaged breakdown needs to be fixed.

In this MR I re-write the matching as suggested by the paper: we make sure every track in a candidate matches a track in the MC signal candidate, rather than calculating an average match with all the LHCbIDs of the entire candidate in one container. This should then make the algorithm able to match effectively a candidate of any number of tracks.

With 1 and 2-track lines, I've observed no change in the performance of the algorithm in terms of the TOS efficiencies it gives on B_s \to J/\psi \phi, which is expected.

TODO

This needs to be tested on a 3 or 4 track line in HLT2, but that requires some work/investigation. The stumbling block is that HLT2 doesn't write out SelReports, which are required to access the LHCbIDs of the trigger candidate. Some work is needed to instead get the LHCbIDs from the turbo-persisted candidates themselves.

Edited by Ross John Hunter

Merge request reports