Best estimate of unfitted seed-track uncertainties for calo-track matching
Seed tracks are used in the calo-track matching (among other track types) to separate neutral and charged clusters. In particular, the x and y position and corresponding uncertainty of the track state at the ECAL plane, propagated from the closest state, are used.
In the baseline configuration, the uncertainty of the track parameters comes from the Kalman Filter and is deemed to be the best estimate. In the light_reco
tracking configuration, seed tracks are not fitted and unfitted seed tracks are used in the calo-track matching. In this case, the track uncertainty comes from the converter fromPrTracksV1Track
. This uncertainty is on purpose relatively large not to bias the Kalman Fit. We need to understand if this uncertainty is suited to be used in the calo-track matching or a dedicated parametrisation can give a better estimate.
Currently (as of !2470 (merged)) seed tracks (before the Kalman fit) use the following uncertainties:
- x^2: 4 (mm^2?)
- y^2: 400 (mm^2?)
- tx^2: 6e-5
- ty^2: 1e-4
- QOverP^2 = (0.15 x QOverP)^2
(see https://gitlab.cern.ch/lhcb/Rec/-/blob/62f9042534ad1635b664706d6a7b06f1dc69bb0e/Pr/PrConverters/src/fromPrTracksV1Track.cpp#L329-337 , https://gitlab.cern.ch/lhcb/Rec/-/blob/62f9042534ad1635b664706d6a7b06f1dc69bb0e/Pr/PrConverters/src/fromPrTracksV1Track.cpp#L262-268 and https://gitlab.cern.ch/lhcb/Rec/-/blob/62f9042534ad1635b664706d6a7b06f1dc69bb0e/Pr/PrConverters/src/fromPrTracksV1Track.cpp#L224 )
Related to Moore!900 (merged) where this discussion was started. See here for some checks of the impact of different seed track uncertainties on the classification of clusters as neutral and charged. The new SelectiveTrackMatchAlg
is much less affected than the classic FuturePhotonMatchAlg
. See here for instructions on how to check the resolution of tracks at the ECAL plane.