Nuclear Physics B Proceedings Supplement 00 (2014) 1–3
Nuclear Physics B Proceedings Supplement
Electron and Photon Reconstruction with the ATLAS Detector
Jovan Mitrevski, on behalf of the ATLAS Collaboration
Fakult¨at f¨ur Physik, Ludwig-Maximilians-Universit¨at M¨unchen, M¨unchen, Germany
Abstract
Excellent performance in the reconstruction of electrons and photons with the ATLAS detector at the LHC is a key requirement for realizing the full physics potential of ATLAS, both in searches for new physics and in precision measurements. For instance, the good electron and photon reconstruction performance played a critical role in the discovery of a Higgs boson, announced by the ATLAS Collaboration in 2012, and in the measurement of its properties.
This paper highlights the reconstruction of electrons and photons.
Keywords: ATLAS, LHC, electrons, photons, reconstruction software
1. Introduction
The performance of electron and photon reconstruc- tion plays a critical role in the reach of many analyses, including H → γγ, H → 4`, beyond the Standard Model searches, and precision measurements. For the 2012 data-taking period of the LHC with√
s=8 TeV, a number of improvements were made to the reconstruc- tion algorithms to increase the efficiency for electrons with low transverse energy (ET) and for electrons and photons in high pile-up conditions.
2. ATLAS Detector
The ATLAS detector [1] is a multi-purpose apparatus with a forward-backward symmetric cylindrical geom- etry and nearly 4πsolid angle coverage. Closest to the beamline is the inner detector, consisting of pixel and microstrip trackers covering |η| < 2.51 and a Transi- tion Radiation Tracker (TRT) covering|η| < 2.0. The
Email address:[email protected](Jovan Mitrevski)
1ATLAS uses a right-handed coordinate system with its origin at the nominal interaction point (IP) in the centre of the detector and thez-axis along the beam pipe. Thex-axis points from the IP to the centre of the LHC ring, and they-axis points upward. Cylindrical co- ordinates (r, φ) are used in the transverse plane,φbeing the azimuthal angle around the beam pipe. The pseudorapidity is defined in terms of the polar angleθasη=−ln tan(θ/2).
TRT uses straw tubes for both tracking and to provide discrimination between electrons and charged hadrons based on transition radiation, which comes from scintil- lating foils and fibres between the straws. The inner de- tector is located inside a thin superconducting solenoid that provides a 2 T magnetic field.
Outside the solenoid, a fine-granularity lead/liquid- argon (LAr) electromagnetic (EM) calorimeter mea- sures the energy and position of electrons and photons in the region |η| < 3.2. In the region |η| < 2.5, the EM calorimeter is segmented into three layers in depth.
The second layer, in which most of the EM shower en- ergy is deposited, is divided into cells of granularity of
∆η×∆φ=0.025×0.025. The first layer is segmented with finer granularity inηto provide discrimination be- tween single photons and overlapping photons coming from the decays of neutral mesons. A presampler, cov- ering |η| < 1.8, is used to correct for energy lost by particles before entering the EM calorimeter.
An iron/scintillating-tile hadronic calorimeter covers the region |η| < 1.7, while a LAr hadronic end-cap calorimeter covers 1.5 <|η| < 3.2. In the forward re- gion, 3.2 < |η| < 4.9, LAr calorimeters with copper and tungsten absorbers measure both the electromag- netic and hadronic energy. A muon spectrometer sur- rounds the calorimeter system.
J. Mitrevski/Nuclear Physics B Proceedings Supplement 00 (2014) 1–3 2
3. Electron and Photon Reconstruction
A number of improvements were made to the electron and photon reconstruction for the 2012 data-taking pe- riod. In this section, we outline the standard reconstruc- tion. For more information on electron reconstruction, see Ref. [2].
3.1. EM Cluster Building
Standard electron and photon reconstruction starts by building clusters out of the energy deposits in the EM calorimeter. The EM calorimeter, using all three layers in depth, is divided into towers of∆η×∆φ = 0.025×0.025. A sliding-window algorithm [3] with windows of size 3×5 inη-φspace is performed, fol- lowed by duplicate removal. Based on MC simulations, the cluster building efficiency is 95 % for electrons with ET=7 GeV, 99 % for electrons withET=15 GeV, and 99.9 % for electrons withET=45 GeV.
3.2. Tracking for Electrons and Photons
EM clusters that pass loose shower shape require- ments in hadronic leakage and energy distribution inη are used to create Regions of Interests (ROIs). Within these ROIs, the tracking is modified as follows.
Standard track pattern reconstruction [4] is first per- formed everywhere, using a pion hypothesis. If the patter recognitions fails for a silicon track seed that is within an ROI, a modified pattern reconstruction algo- rithm, based on a Kalman filter formalism [5], using an electron hypothesis and allowing for up to 30 % energy loss at each material surface, is performed. Track can- didates are then fitted with the globalχ2 fitter [6], ini- tially using the same particle hypothesis as was used in the pattern recognition, but retrying with an electron hy- pothesis if the original pion hypothesis fails.
Tighter quality requirements have been made to im- prove the track purity, especially for TRT stand-alone (TRTSA) tracks. Matching of the tracks to the EM clus- ters is then performed. A tracks is considered loosely matched to a cluster if either (i) when extrapolated to the second layer of the EM calorimeter it is near the cluster inφ, and if it has silicon hits, inη, or (ii) as above, but the track momentum is rescaled to the energy measured in the EM cluster when performing the extrapolation.
Tracks with silicon hits loosely matched to EM clus- ters are refitted with a Gaussian Sum Filter (GSF) fit- ter [7], a non-linear generalization of the Kalman fil- ter, for improved track parameters, as demonstrated by Fig. 1. These tracks, along with the TRTSA tracks, are used for track matching and conversion vertex building.
truth
(q/p)
truth
-(q/p) (q/p)reco
-0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Entries / 0.01875
0 100 200 300 400 500 600
103
×
ATLASPreliminary Simulation
= 7 TeV s
ee (Standard) Z→
ee (GSF)
→ Z
truth
(q/p)
truth
-(q/p) (q/p)reco
-0.4 -0.2 0 0.2 0.4 0.6 0.8 1
GSF/Standard
2 4 6 8 10
Figure 1: The relative bias inq/p[7].
3.3. Track Matching for Electrons
The loosely-matched tracks are matched to EM clus- ters with slightly tighter requirements in η andφ. At least one track must be matched for a cluster to form an electron. If multiple tracks are matched, they are sorted in the following order, with the first being used for the electron properties. Preferred are tracks with hits in the pixel detector, then those with silicon hits but no pixel hits, then TRTSA tracks. Within each category, those that have a better∆Rmatch inη-φ, where the extrapo- lation is done with the track momentum rescaled to the cluster energy first and unrescaled second, are preferred, unless the differences are small, in which case the one with more pixel hits is preferred, giving an extra weight to a hit in the innermost layer.
3.4. Photon Conversion Reconstruction
Conversion-finding is run on the loosely-matched tracks. Converted photons are classified as single-track or double-track. Double-track conversions are created when two tracks form a vertex consistent with coming from a massless particle. Single-track conversions are essentially tracks not having hits in the innermost sen- sitive layers. To increase the purity, the tracks used to build conversions must generally have a high probabil- ity to be electron tracks as determined by the TRT, espe- cially when building single-track conversions or if using TRTSA tracks. The matching between tracks from the conversion vertices and clusters was tightened for the 8 TeV running. Quality cuts are applied on the conver- sion vertices for better pile-up tolerance. Because dead pixels can effect conversion building, the dead pixel map is also used to determine the conversion quality. If
J. Mitrevski/Nuclear Physics B Proceedings Supplement 00 (2014) 1–3 3
there are multiple conversion vertices matched to a clus- ter, double-track conversions with two silicon tracks are preferred over other double-track conversions, followed by single-track conversions. Within each category, the vertex with the smallest conversion radius is preferred.
3.5. Final Cluster Creation
At the end, new, calibrated EM clusters are created for electrons and photons. In the barrel, the clusters are of size 3×5 (3×7) inη-φspace for unconverted photons (electrons and converted photons), while in the end-caps, the clusters are of size 5×5, measured in second-layer cells.
4. Results and Conclusion
As can be seen in Fig. 2 and Fig. 3, significant im- provements in the reconstruction of electrons have been made for the 8 TeV running, especially for electrons at lowET and at higher pseudorapidity, where the tracks traverse more material and hence are more likely to lose energy due to bremsstrahlung. The figures show the ef- ficiency to find, fit, and match a track to a calorimeter cluster. The efficiency was increased by 5 % overall, 7 % for low-ETelectrons. This improvement has had a significant effect on many analyses, includingH→4`.
η -2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Efficiency (Reco)
0.8 0.85 0.9 0.95 1 1.05
: 15-50 GeV ET
L dt = 4.7 fb-1
∫
=7 TeV s 2011 data 2011 MC
L dt = 20.7 fb-1
∫
=8 TeV s 2012 data 2012 MC
ATLAS Preliminary
Figure 2: Electron reconstruction efficiency as a function ofη[2].
For photons, the main goal was to improve the pu- rity of the conversions, especially at high pile-up. Fig- ure 4 shows a stable behavior of the photon reconstruc- tion as a function of the average number of interactions per bunch crossing. Without the updates, the number of conversions would have increased significantly at high pile-up, indicating a contribution from fake conversions.
This improvement has had a significant effect on many analyses, includingH→γγand SUSY searches.
In summary, the improvements that were made to the electron and photon reconstruction software have been
[GeV]
ET
20 30 40 50 60 70 80
Efficiency (Reco)
0.75 0.8 0.85 0.9 0.95 1 1.05
<2.47
η
L dt = 4.7 fb-1
∫
=7 TeV s 2011 data 2011 MC
L dt = 20.7 fb-1
∫
=8 TeV s 2012 data 2012 MC ATLAS Preliminary
Figure 3: Electron reconstruction efficiency as a function ofET[2].
Average interactions per bunch crossing
5 10 15 20 25 30 35
Fraction of photon candidates
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Unconverted photons Converted photons
Single track conversions Double track conversions
ATLAS Preliminary = 8 TeV s Data 2012,
L dt = 3.3 fb-1
∫
Figure 4: The fraction of photon candidates that are reconstructed as unconverted, single-track conversions, or double-track conversions.
very important in achieving the many physics goals of Run I at ATLAS.
References
[1] ATLAS Collaboration, The ATLAS Experiment at the CERN Large Hadron Collider, JINST 3 (2008) S08003.
doi:10.1088/1748-0221/3/08/S08003.
[2] ATLAS Collaboration, Electron efficiency measurements with the ATLAS detector using the 2012 LHC proton-proton collision data, Tech. Rep. ATLAS-CONF-2014-032, CERN, Geneva (Jun 2014).
[3] W. Lampl, et al., Calorimeter clustering algorithms: descrip- tion and performance, Tech. Rep. ATL-LARG-PUB-2008-002, CERN, Geneva (Apr 2008).
[4] T. Cornelissen, et al., Concepts, design and implementation of the ATLAS New Tracking (NEWT), Tech. Rep. ATL-SOFT-PUB- 2007-007, CERN, Geneva (Mar 2007).
[5] Fr¨uhwirth, R., Application of Kalman filtering to track and vertex fitting, Nucl.Instrum.Meth. A262 (1987) 444–450.
doi:10.1016/0168-9002(87)90887-4.
[6] T. G. Cornelissen, et al., The global χ2 track fitter in AT- LAS, J.Phys.Conf.Ser. 119 (2008) 032013. doi:10.1088/1742- 6596/119/3/032013.
[7] ATLAS Collaboration, Improved electron reconstruction in ATLAS using the Gaussian Sum Filter-based model for bremsstrahlung, Tech. Rep. ATLAS-CONF-2012-047, CERN, Geneva (May 2012).