Noise dependence with pile-up in the ATLAS Tile calorimeter
J. P. Araque !on behalf of the ATLAS Tile Calorimeter System
ANIMMA 2014 Lisbon, 20-24 April 2015
The Tile Calorimeter
• The Tile Calorimeter is the central hadronic calorimeter of the ATLAS detector.
• Composed of iron layers (as passive material) and scintillating plastic layers (as active material).
• Structure: • 4 partitions (LBA,LBC,EBA and EBC). • 4 layers (A,BC,D and special layers). • 16 towers (steps of 0.1 in η). • 64 modules (divisions in the azimuthal
angle φ).
2
Pile-up in the ATLAS detector
• Tens of proton-proton interactions take place in a bunch crossing inside the ATLAS detector.
• Pile-up events also deposit energy in the TileCal cells which is hard to distinguish from the energy deposited by the interesting event.
• This is known as pile-up noise and is a component to the cell noise measured that increases as pile-up increases.
3
-400 -200 0 200 400 600
Nor
mal
ized
ent
ries
-310
-210
-110
ATLAS PreliminaryTile calorimeterEBA Cell A12
=58.6 MeVσ>=20, µMC12 <=89.5 MeVσ>=30, µMC12 <
=80.8 MeVσ>=20, µData 2012 (50 ns) <=97.3 MeVσ>=30, µData 2012 (50 ns) <
Energy [MeV]-400 -200 0 200 400 600
Dat
a/M
C
0.51
1.52
2.5 >=20µ>=20 / MC <µData <>=30µ>=30 / MC <µData <
Mean Number of Interactions per Crossing0 5 10 15 20 25 30 35 40 45
/0.1
]-1
Rec
orde
d Lu
min
osity
[pb
020406080
100120140160180 Online LuminosityATLAS
> = 20.7µ, <-1Ldt = 21.7 fb0 = 8 TeV, s
> = 9.1µ, <-1Ldt = 5.2 fb0 = 7 TeV, s
RMS as the noise estimator
• The RMS of the energy distribution can be used as a noise estimator:
4
phE2i � hEi2
>µ<0 50 100 150 200
Noi
se [M
eV]
0
50
100
150
200
250 ATLAS PreliminaryTile Calorimeter
= 8 TeVs50 ns, Layer A
DataMonte Carlosimulation
• The noise measured increases with the mean number of interactions per bunch-crossing.
• The bunch-spacing also increases the cell noise.
Quantiles of the energy distribution as the noise estimator
• The quantiles of the energy distribution can be used to better characterise its shape.
• The energy ε is the kth quantile of the energy distribution with Q quantiles if:
5
P (E ✏) = k/Q.
>µ<10 15 20 25 30 35
E [M
eV]
-2000
-1000
0
1000
2000
3000
4000
5000
600099.99 % of events99.73 % of events95.45 % of events68.27 % of events
ATLAS PreliminaryTile Calorimeter
= 8 TeVsData 50 ns, | < 0.7ηLayer A, 0.6 < |
• Non-gaussian behaviour with larger positive tails.
>µ<10 15 20 25 30 35
Noi
se [M
eV]
210
310
410
51099.99 % of events
RMS×495.45 % of events
RMS×2
ATLAS PreliminaryTile Calorimeter
= 8 TeVsData 50 ns, | < 0.7ηLayer A, 0.6 < |
>µ<10 15 20 25 30 35
Noi
se [M
eV]
210
310
410
510 99.99 % of eventsRMS×4
95.45 % of eventsRMS×2
ATLAS PreliminaryTile Calorimeter
= 8 TeVsData 50 ns, | < 0.8ηLayer D, 0.6 < |
Conclusions
6J.P. Araque funded by LIP and by the FCT grant SFRH/BD/52002/2012
• The high rate at which the LHC produces proton-proton collisions implies a high amount of pile-up events taking place in the ATLAS detector.
• In the second operation phase of the LHC, with higher energy (13 TeV) and smaller bunch-spacing (25 ns), pile-up is expected to increase.
• To study the TileCal cell noise as a function of pile-up two estimators have been presented:
• The RMS of the energy distribution.
• The quantiles of the energy distribution.
• Using the quantiles estimator the shape of the energy distribution can be better characterised.