Old-new Track Event Model Comparison studies
Reminder of the definitions of the variables plotted:
- OT measurement = calibrated time from OTTime x ambiguity (in ns)
- error on OT measurement = 2.6 ns, corresponding to a 200 μm resolution:
from the XML definitions one gets
errMeas = resolution * driftDelay = (200 μm) * (32.5 ns/2.5 mm) = 2.6 ns
- VELO phi measurements = distance to strip -> either 2.8 or 3.1 mm (signed number)
- IT measurement = coordinate measured in mm = channelID converted to position + distance to strip
- error on IT measurement in mm
- chi2 = (res/errRes)2
Old ideal pattern recognition output converted to new Tracks
(05/07/2005)
The plots were produced in the following way:
- old Track Event Model (TEM) ideal pattern recognition with creation of true states at all measurement positions
-> production of TrFitTracks
- there was no selection on the quality of the tracks (momentum, number of hits, etc.)!
- conversion to new Tracks with creation of LHCbIDs from old measurements
and then creation of the new Measurements from the LHCbIDs
- all the pairs (state,measurement) were looped over, and the chi2 and res/errMeas plots were
produced via the TrackMasterProjector that takes care of using the correct OT/IT/Velo-R/Velo/-phi
projector
First look at performance with the old track event model
(30/06/2005)
The plots were produced in the following way:
- old Track Event Model (TEM) ideal pattern recognition with creation of true states at all measurement positions
-> production of TrFitTracks
- there was no selection on the quality of the tracks (momentum, number of hits, etc.)
-> first look including also low-quality tracks
- conversion to new Tracks with creation of LHCbIDs from old measurements
and then creation of the new Measurements from the LHCbIDs
- all the pairs (state,measurement) were looped over, and the chi2 and
(residual/(error on measurement) plots were produced via the TrackMasterProjector
that takes care of using the correct OT/IT/Velo-R/Velo/-phi projector
In the VELO plots below both the R and phi measurements have been histogrammed together.
In the VELO measurement plot("Velo Meas") there are clearly 2 parts: phi-measurements in the first bins
and R-measurements in the region 10-40. The phi-measurements are signed and can be either +/-2.8 or +/-3.1mm.
There are also 2 regions (the first is very much the first bins) for the histogram of the error on the
VELO measurements ("Velo errMeas").
Some points remain to be understood, though it is clear that one needs to repeat this exercise with
a rather clean sample of tracks, applying quality cuts on momentum, numbers of hits, etc.:
The main worry refers to the Gaussian fits of the residual/error-on-measurement plots for the IT and Velo,
which give a sigma significantly away from 1.