E/p Cut Efficiencies
We cut on E/p for our candidate daughter electrons and positrons for
the AuAu analysis. To find the efficiencies of these cuts, we find
the efficiency of the cuts on embedded single electrons from the
file AuAu2010_singleE_emb_4May12.root, located in /home/khill/upsAuAu/ on
the nuclear server, appropriately modified to match the data, as described
below. We used data from the file AuAu10_ups_10Jun12.root, located in
/home/kesich/AuAu2010_Upsilons/ on the nuclear server.
For the first, L0, daughter from the data we select only positively charged
particles with nSigmaElectron > 1 and with cluster energies between 5 and
6. We compared these to embedded electrons with Adc > 303 and cluster
energy between 5 and 6. They also must have been reconstructed in the tpc
and have been from an event with at least one fired calorimeter tower. We
plotted both these real particles' and embedded electrons' Ecluster/p
spectrum. We fit the data around the peak (from 0.7 to 1.5) and the
embedding in the full range to gaussians.
However, the embedded particles were modified to match the data. Firstly,
since the embedding was thrown flat in momentum, we weighed the momentum
spectrum to the data. Secondly, we smeared the Ecluster values to better
simulate the resolution of the calorimeter. Thirdly, we still found the
mean of this modified electron embedding signal to not fall on top of the
peak in the data. Realizing that contamination from other particles in
the event can add to the measured energy from the calorimeter, we felt
justified in simply shifting the mean of the electron embedding to match
this peak. To instantiate these last two modifications, we convolved the
embedding E/p spectrum with another gaussian, iteratively tweaking the mean
and width of the convolving gaussian until the embedding's and data's
gaussian fits matched.
For the second daughter from the data, we again select only positively
charged particles with nSigmaElectron > 1. We found that the second
daughters' cluster energies were not well correlated to their momentum,
indicating contamination from particles unrelated to Upsilon production.
Note that the energy requirement of the L0 trigger on the first daughter
probably ensures the much cleaner correlation of momentum and cluster
energy for the first daughters. Tower energies did, however, show close
equality to momentum values in the second daughter. Cutting on tower
energy between 5 and 6 provided inadequate statistics so we cut instead on
tower energy from 3 to 4. The embedded electrons were cut with Adc > 303,
as for the comparison with the first daughters, but now necessarily with
tower energy between 3 and 4 for comparison to the second daughters. We
fit the data around the peak (now, from 0.7 to 1.4) and the embedding in
the full range to gaussians, like for the first daughters.
A similar process for analysis of the first daughters was done with the
second daughters to match embedding to the data; the embeddding momentum
spectrum was weighed to data and a gaussian was convolved over the
embedding, as described above.
For the first daughter, we cut on Ecluster/p between 0.7 and 1.4.
One can calculate the resultant efficiency by considering our matched
embedding distribution in two ways. Firstly, we can integrate the
gaussian fit to the embedding between the cuts and divide by its integral
over all E/p. Secondly, we can integrate the histrogram between the cuts
and divide by the histograms integral over all E/p. Using the fit, the
efficiency of the first daughter's cut is calculated to be ~90%.
Using the histogram, the efficiency is calculated to be ~88%.
For the second daughter, we cut on Etower/p between 0.7 and 1.4. We can
calculate the efficiency, again, in two ways. Using the fit, the efficiency
is calculated to be ~87%. Using the histogram, the efficiency is ~84%.