Presentation is loading. Please wait.

Presentation is loading. Please wait.

Performance of jets algorithms in ATLAS

Similar presentations


Presentation on theme: "Performance of jets algorithms in ATLAS"— Presentation transcript:

1 Performance of jets algorithms in ATLAS
Nikola Makovec September the 8th, 2009

2 Experimental Requirements for Jet Finders
Detector independence: The performance of the algorithm should be as independent as possible of the detector that provides the data. For example, the algorithm should not be strongly dependent on detector segmentation, energy response, or resolution. Minimization of resolution smearing and angle biases: The algorithm should not amplify the inevitable effects of resolution smearing and angle biases. Stability with luminosity: Jet finding should not be strongly affected by multiple hard scatterings at high beam luminosities. For example, jets should not grow to excessively large sizes due to additional interactions. Furthermore the jet angular and energy resolutions should not depend strongly on luminosity. Efficient use of computer resources: The jet algorithm should provide jet identification with a minimum of computer time. However, changes in the algorithm intended to minimize the necessary computer resources, e.g., the use of seeds and preclustering, can lead to problems in the comparison with theory. In general, it is better to invest in more computer resources instead of distorting the definition of the algorithm. Maximal reconstruction efficiency: The jet algorithm should efficiently identify all physically interesting jets (i.e., jets arising from the energetic partons described by perturbative QCD). Ease of calibration: The algorithm should not present obstacles to the reliable calibration of the final kinematic properties of the jet. Ease of use: The algorithm should be straightforward to implement with typical experimental detectors and data. Fully specified: Finally, the algorithm must be fully specified. This includes specifications for clustering, energy and angle definition, and all details of jet splitting and merging. Blazey et al. hep-ex/000501 Nikola Makovec

3 The Atlas Calorimeters
EM Endcap EMEC EM Barrel EMB Hadronic Endcap Forward Tile Barrel Tile Extended Barrel Electromagnetic Barrel || < 1.4 Electromagnetic EndCap 1.375 < || < 3.2 Hadronic Tile || < 1.7 Hadronic EndCap 1.5 < || < 3.2 Forward Calorimeter 3.2 < || < 4.9 At least 10 interaction lengths ~ channels Varied granularity Varied technologies Overlap/crack regions Nikola Makovec

4 Calorimeter Signals: Towers
Regular grid Δη×Δφ = 0.1×0.1 Calorimeter cell signals are summed up in tower bins No cell selection, all cells are included Indiscriminatory signal sum includes cells without any true signal at all Sum typically includes geometrical weight Towers have fixed direction given by geometrical grid center Massless four-momentum representation projective cells non-projective cells Nikola Makovec

5 Calorimeter Signals: Topological clusters
Attempt reconstruction of individual particle showers Uses cell signal significance Electronic Electronic + pile-up (quadratic sum) Attempt to suppress noise with least bias on physics signals Algorithm: Seeding (S=4) Cells with signal significance above primary seed threshold Collecting (P=2) Directly neighboring cells with signals above basic threshold, directly neighboring seed cells in 3-dim Growth control (N=0) Collect neighbors of neighbors if those have signals above secondary seed significance Splitting Find local signal maxima in cluster with E > 500 MeV Split cluster between those Nikola Makovec

6 Calorimeter Signals: TopoTowers
Noise suppressed towers Electromagnetic energy scale signal only Only cells surviving noise suppression are used in the towers TopoClusters are used as noise suppression tool Additional cell filters possible Nikola Makovec

7 Degrees of freedom Jet algorithm: Inputs: Size: ATLAS seeded cone
SISCone Kt AntiKt Inputs: Truth particles (excluding neutrinos and muons) Towers Topocluster Topotower Size: Narrow R=0.4 Wide R=0.6 or 0.7 Nikola Makovec

8 Timing: HLT Nikola Makovec

9 Timing: offline Nikola Makovec

10 Energy scale and resolution
Nikola Makovec

11 Efficiency Nikola Makovec

12 Purity Nikola Makovec

13 Pile-up: multiplicity
Nikola Makovec

14 Pile-up: multiplicity
Nikola Makovec

15 Pile-up: constituents
Nikola Makovec

16 Pile-up: response Nikola Makovec

17 Randomly triggered events
Nikola Makovec

18 Randomly triggered events
Nikola Makovec

19 Randomly triggered events
Nikola Makovec

20 Back up Nikola Makovec

21 Efficiency The plots are done using QCD MC events [560-1120GeV]
Nikola Makovec

22 Nikola Makovec

23 Pile-up: response Nikola Makovec

24 Pile-up: response Nikola Makovec

25 75% Et cut on proto-jet Dark tower Nikola Makovec


Download ppt "Performance of jets algorithms in ATLAS"

Similar presentations


Ads by Google