TSF data/MC comparison study
Further study on the hit
efficiency suggested to look at
Ghit charge distributions and use it as a tool to adjust
the efficiency.
The following plots show a
comparison between data and MC using Bhabha event and run16802.
The data charge was scaled by a factor
of 11.8 and one plot was normalized to the other.
As we see from the previous plots
the Ghit charge distribution is narrower than the data.
This is an expected Landau charge distribution in thin absorber or gas.
In order to simulate the right charge distribution that include
for example electronic noise,
I applied Gaussian smearing to the Ghit charge.
The amount of smearing was studied by comparing the Ghit
charge distribution from flat bhabha events from MC (generated with 8.8.0g)
and data run 16802.
Eventually 30% smearing was applied in order to have a match as can be seen in the following plots.
To make sure the charge distribution
well describe the physics I looked at few different momentum ranges
for a particular path length range.
The next plots show charge and charge divided by the path length distributions
for hadronic events with path length between 1 < path < 2 cm but for different monentum ranges.
I study different momentum regions
of the dE/dx where we expect different beta dependence
in order to test the smeared charge distribution
The first two plots are in momentum range of 0.1 < P <0.15 GeV
The following plots are in momentum range 0.3 < P < 0.9 GeV
The following plots are in momentum range 2. < P < 6. GeV
In addition I have looked at Bhabha events with very long path length 3 < path < 4 cm .
In this case there not too many events
I compared data and MC for momentum rage of 2. < P <6. GeV
The last one is for path length
between 2 < path < 3 cm and mometum
between 0.1 < P < 5.GeV
The Gaussian smearing allowed us to have a resolution for particles in agreement with true data,
especially for low charge
particles that we are going to use as our parameter to reject
hit.
The charge smearing was added
as section criteria in trgDC with 5 new tcl parameters
to adjust the hit efficiency.
In the following I have
done TSF data/MC comparison selecting hadronic events,
the new hit efficiency is shown for data 16802 and MC generated
with 8.8.0h
The new scaling for the lower voltage data is 6.86
The following plots are
comparison between data/MC selecting hadronic events
for data 12396
and the MC bbgeneric generated with 8.8.0h
the hit efficiency
It looks like smearing the charge
and adjusting the efficiency depending on the charge
only is not sufficient.
The reasons might be:
- The charge smearing cuased the lost of the theta dependency information
since the smearing was too.
- The cut should have path length and beta dependency
- Smearing differently for different path length gives reasnable
charge distribution
bhabha data-MC charge distribution.
but it was not tested as a cut in trgDC yet
An interesting point to check is
Is there any gain differences between the different SL
In the following we have profile histogram of the charge for every super layer (run 16802)
We dont see any large differences in gain for the different super layer
and the changes vesus phi seem statistical.
Since we saw that the efficiency to get a good segment coming close to
the interaction point and going stright up in the MC
is better than in the data.
The question is there gain differences
between the different layers that
is not simulated in the MC
and cuase a inefficiencies to weight
equal to 3 segments. (run 16802)
The answer to this question is selfexplanetory in the next plot studied with run 16802
For run 12396 which had lower voltage
|
