for Moore!3221 Rec!3833
Timothy David Evans (98521896) at 27 Mar 10:22
Merge branch 'ausachov_cheated_track_fit' into 'run5'
... and 1 more commit
Adding cheated track fit with
qop
value and add option to add ideal states to the trackTrackResCheckerNT
This is invasive, and I am happy to address any comments. Goes with Moore!3221 LHCb!4499
Timothy David Evans (b70b409b) at 27 Mar 10:11
Merge branch 'ausachov_cheated_track_fit' into 'run5'
... and 1 more commit
Adding cheated track fit configuration with
TrackResolutionChecker
- both nTuple and Histo versionsAs for now, Long fit runs fine. Upstream and Downstream fits return errors during the track fit sometimes.
Timothy David Evans (20da0f9b) at 27 Mar 10:10
Merge branch 'ausachov_cheated_track_fit' into 'run5'
... and 1 more commit
Adding cheated track fit configuration with
TrackResolutionChecker
- both nTuple and Histo versionsAs for now, Long fit runs fine. Upstream and Downstream fits return errors during the track fit sometimes.
Okay, for the first case it sounds like any HLT1 configuration works fine, in which case it is most efficient/reproducible just to run the filtering using the most recent Allen/Moore release + appropriate DD4Hep settings.
For case (2): to me it makes sense to separate concerns, meaning keeping the tests running on DetDesc until we can produce simulated samples with DD4Hep (otherwise we are mixing the bandwidth test with what is essentially testing the interoperability of DetDesc samples with DD4Hep, which is of course very important, but also very separate from a test of the HLT2 bandwidth)
These MDFs are produced using the 'standard' HLT1 production (i.e. allen_hlt1_production) in Moore.
They were produced using DetDesc rather than DD4Hep, with tags matching those used in the MC production itself.
Depending on what the purpose of this test is tells us how to proceed,
Timothy David Evans (a8782a2a) at 26 Mar 10:20
remove unneccessary include of odin in the validators
The v2 samples were basically produced with the 1MHz settings (modulo some changes not yet merged to master for the Lambda lines, and some different prescales on technical lines)
The next update of these samples will be when we have the lines tuned on data taken during the ramp.
python linting sucks. But indeed this can be removed.
Timothy David Evans (8e5d7f38) at 26 Mar 10:08
Fixed formatting
Timothy David Evans (cb472159) at 26 Mar 10:06
apply correct prescale to Jpsi alignment line
... and 1 more commit
Done
Done
Done
Tuning means insanely loose thresholds that you can then only tighten to get to a reasonable rate. Its things like setting the MVA cut to 0 on the TwoTrackMVA.
I don't know if you have a suggestion for better name?
Done