1. First attempt to apply a calibration on real data Iacopo Vivarelli, Andrea Dotti, Chiara Roda

2. Introduction
The aim of this talk is to try to get a feeling on the expected degradation of performance of a calibration once we move from MonteCarlo to data.
The calibration discussed yesterday by Andrea (so called “Pisa calibration”) has been tuned for jets reconstructed for TopoClusters, cone 0.7
It has been shown that it works fine if the cone is reduced to 0.4, which means that the dependence on the lateral jet shower shape is low
In order to asses how well the calibration obtained from MC will work on data we can use CTB data. The best method would be: get weights from CTB MC and apply them to the data.
As a first try: we applied “blindly” the calibration to test beam pions (first MonteCarlo, then data). The relevant information is the relative difference of MC Vs data

3. CTB pions
It is only the first glance. We did not pay too much attention to the event selection. Probably it can be improved.
We looked at two energy points: 50 GeV and 180 GeV
For data: runs and
MonteCarlo: official centralized production v11.0.3a
EVENT SELECTION
Only for the data: request of physics trigger (Trigger = 1) and beam purity (cuts on the beam chambers).
Muon removal: a signal in the MuTag scintillator or a signal compatible with a muon in all the three samples of TileCal.

4. CTB pions(2)
Electron removal: look at the distribution of LAr EMB1 VS LAr EMB2, make a projection, then make a cut.

5. Pion energy
Pions are reconstructed as jets, then the “Pisa” calibration is applied
Both in MC and in the data, some linearity is recovered
High energy tails

6. Pion energy(2)
The high energy tail is caused mainly by pion which are MIPs in the LAr
We could expect the calibration not to calibrate those pions  the weights have been computed on jets  basically no jets are MIPs in LAr

7. Large improvement for the resolution in MC. Small (if any) on data
Results at 50 GeV
Let us make a crude cut (we are not interested in a precise measurement of the pion spectrum). We reject the event if ELAr/Ejet < 0.15.
The mean values alone are nice, but not necessary relevant. What is important here is the relative difference between the two (almost 4%)
Large improvement for the resolution in MC. Small (if any) on data
Mean = GeV
Mean = GeV

8. Results at 180 GeV
At 180 GeV, the relative difference on the mean value is about 2%
There is some improvement in the resolution also on data
Mean = GeV
Mean = GeV

9. Conclusions
We did a (quite rough) analysis of 50 GeV and 180 GeV data from CTB 2004, extracting pions. We also used simulated 50 GeV and 180 GeV pions
We reconstructed the pions as jets. We applied the “Pisa” calibration out of the box.
50 GeV: the linearity is recovered at 2% level for both MC and data. The most relevant information: the relative difference between MC and data is about 4%. Large resolution improvement for MonteCarlo, almost none for data
180 GeV: the linearity is recovered at maximum 4%. Relative difference MC-DATA = 2%. Improvement in the resolution more relevant for data.