20-Dec-04 F. Merritt 1 Jet Reconstruction and Calibration in ATLAS Frank Merritt / Ambreesh Gupta...
-
Upload
milo-smith -
Category
Documents
-
view
218 -
download
0
Transcript of 20-Dec-04 F. Merritt 1 Jet Reconstruction and Calibration in ATLAS Frank Merritt / Ambreesh Gupta...
20-Dec-04F. Merritt
1
Jet Reconstruction and Calibration in ATLAS
Frank Merritt / Ambreesh GuptaUniversity of Chicago
North American Atlas Physics ConferenceTucson, Arizona
December 20, 2004
(version 3.0)
20-Dec-04F. Merritt
2
JetEtMiss Group
Co-conveners: D. Cavalli, A. Gupta, F. Merritt
Work reported here:• Kyle Cranmer, Ambreesh Gupta, Peter Loch, Sanjay Padhi, Frank Paige
Other U.S./N.A. Contributors:• A. Farbin, J.Proudfoot, T. LeCompte, M. Hurwitz, S. Rajagopolan, Irene Vichou; M.Weilers, R. Mazini
Key European participants:Donatella Cavalli, Sylvia Resconte, Chira Roda, Iocopo Vivarelli, Martine Bosman, Caroline Deluca, Sasha Solodkov, Dan Tovey, David Rousseau
20-Dec-04F. Merritt
3
Main areas of activity in 2004
•Jet-Finding Algorithms:–Cone–Seedless cone–Kt
•Integration of TileCell, LArCell into CaloCell, etc.•Extensive changes in response to RTF recommendations on jet classes, navigation.•New Tools•Topological cluster-finding (Sven Mencke)
•Calibration:–BNL (Frank Paige, Sanjay Padhi)–Chicago (Ambreesh Gupta; A. Farbin, F. Merritt)–Pisa (Chiara Roda, Iacopo Vivarelli)
•Extensive work of Hadron Calibration group•Major test-beam effort (data-taking plus software development)
20-Dec-04F. Merritt
4
General algorithm and tool features:General algorithm and tool features:
only one algorithm JetAlgorithm for all jet finding strategies:
owns the tools performing the actual job;
retrieves the input collection (generic INavigable4MomentumCollection) and converts objects in this collection into Jets held in a optimized internal collection;
… but algorithm does not fail if no input data to be retrieved -> special tool can load in special data object collection(s) which cannot be addressed as INavigable4Momentum, like truth particles, for specific internal conversion;
algorithm and tools are protected from specifics of the internal collection (JetUtils/JetCollectionHelper provides all memory management etc.);
extensive use of all kinds of other helpers by algorithm and tools -> avoid duplication of code and ensures coherent behaviour:
JetUtils/JetSorters for up or down sorting by various kinematic variables;
JetUtils/JetDistances for distances in r, eta, and phi (actually uses FourMom helpers);
JetUtils/JetSelectors for jet selection by cuts in kinematic variables ( in/out ranges, above/below thresholds);
From Peter Loch: SW week, May 2004
20-Dec-04F. Merritt
5
Example Kt detector jet finder algorithm flow (8.2.0):Example Kt detector jet finder algorithm flow (8.2.0):
Kt JetAlgorithm
Transient Event Store
JetSignalSelectionTool
JetPreClusterTool
JetKtFinderTool
JetCellCalibTool
JetSignalSelectionTool
convert to Jets
record JetCollectio
n
INavigable4MomentumCollection
JetCollection
symLinked to concrete CaloTowerContainer, CaloClusterContainer, CaloCellContainer, TrackParticleContainer, eflowObjectContainer,…
KtTowerJet_jobOptions.py
20-Dec-04F. Merritt
6
20-Dec-04F. Merritt
7
20-Dec-04F. Merritt
8
Main problem areas
• Calorimetry effects:– Non-compensation of Atlas calorimeters– Cracks and dead material– Boundaries between calorimeters
• Definition of “truth”– Can apply reco algorithms to MC particle list to obtain MC “jets”.
But is this truth? Clustering is different, propagation is different.– Can sum all MC particles in cone around reco jet.
• Noise. – Want to reject cells with no real energy,but…– Also want to avoid bias. Reject E<0 => +300 GeV bias per event!
20-Dec-04F. Merritt
9
20-Dec-04F. Merritt
10
Three Weighting Schemes Being Studied
• “Pseudo-H1 weighting” [Frank Paige]– Estimates weight for each CaloCell depending on energy density
in cell. Independent of Jet energy.
• Weight by Sampling Layer [Ambreesh Gupta]– Estimates weight for each sampling layer in the calorimeter
depending on Jet energy (but not on cell energy).
• Pisa weights [C. Roda, I. Vivarelli]– Estimates weight for each CaloCell depending on both cell energy
and jet energy (and parameterized in terms of Et rather than E).
20-Dec-04F. Merritt
11
“Pseudo-H1 fitting”(Frank Paige)
• Weight for each cell depends only energy density of that cell and calorimeter type.
• Cells with larger energy are more probable to have a larger EM fraction, so they get smaller weight
• This method allows cell energy to be determined BEFORE jet-finding is executed.
In the figure, blue is energy at EM scale (uncalibrated), red is reconstructed using H1 weights.
20-Dec-04F. Merritt
12
• Resolution is comparable to that of the TDR (0.5/E).
• But no noise
• Ideal cells
• Note: resolution worsens with higher .
Blue is uncalibrated (EM scale) energy
Red is after H1 weights.
Fairly uniform energy calibration over three decades of E.
20-Dec-04F. Merritt
13
Comparison with jet-finding applied to topological clusters:
20-Dec-04F. Merritt
14
Sampling Layers
EM Cal LAr calorimeter
HAD Cal Tile+HCAL+FCAL• No noise added• Calibration weights derived in
three eta region -
0.0 - 0.7, 0.7 - 1.5,
1.5 - 2.5, 2.5 - 3.2• The weights have reasonable
behavior in all eta region.
“Sampling Weights”
(Ambreesh Gupta)
20-Dec-04F. Merritt
15
20-Dec-04F. Merritt
16
20-Dec-04F. Merritt
17
/E = (97% /E) 4%
/E = (127% /E) 0%
/E = (114% /E) 8%
/E = (68% /E) 3%
Scale & Resolution
Sampling Weights
20-Dec-04F. Merritt
18
/E = (75% /E) 1% /E = (115% /E) 3%
/E = (138% /E) 0%/E = (271% /E) 0%
Scale & Resolution
H1 Style Weights
Different definition
of truth, compared
to those used in
deriving the weights
20-Dec-04F. Merritt
19
Jet Calibration/Testing
A Jet calibration scheme specifies the following
1. Definition of truth Jet.
2. Treatment of noise - (a)symmetric cut on CaloCell, higher objects, more sophisticated noise algorithm. . .
3. Input to Jet clustering algorithm - CaloTower, CaloCluster. . .
4. Type of Jet clustering algorithm - cone (0.7,0.4), kt . . .
5. Calibration Weights to be applied to the Jet - H1-Style, H1, Sampling . . .
20-Dec-04F. Merritt
20
Testing a calibration scheme
1. Using the definition of truth specified by the calibration scheme, some basic
quantities can be estimated
- Linearity of Jet energy scale and error
- Change in calibrated Jet resolution (EM scale resolution should be same
for different schemes)
- Effect on symmetry-asymmetry, skewness of energy distribution
- Effect on angle
2. Effect of calibration weights on
- samples with different event topology ttbar, dijet
- samples with different content of quark and gluon jets
3. Jet energy scale can tested independent of truth definition with the ‘in situ’
samples
- Z/ + Jet,W jj in ttbar sample (can be done in both data and MC)
- Hadronic decay of Z in Monte Carlo
- bootstrap to higher energies
Jet Calibration/Testing
20-Dec-04F. Merritt
21
Different groups/methods working in Jet calibration utilise different information
in the detector. This helps in an optimal use of all information and cross checks.
In order to compare results, a simple strategy could be to
1. Apply a symmetric noise cut on CaloCell
2. Choose one Jet Algorithm - cone(0.7)
3. Choose one definition of truth Jet - truth particle within the Jet.
Calibration weights derived with this will not be the optimal one, but it would
be a good strating point to understand differences between algorithms.
Comparing Jet calibration schemes
20-Dec-04F. Merritt
22
Ongoing work and plans for next two months (in preparation for Rome)
1. Pisa wieghts are in the process of being put into JetRec for comparison to H1 and sample weighting.
2. Will introduce a top-level calibration selector tool in JetRec that can be switched through jobOpt.
3. Will carry out comparisons in January with the goal of establishing a benchmark calibration by early February.
4. Produce new DC2 weights by mid-February (already in progress; F.P. and S.P.)5. Extend calibration to different cone sizes (R=0.4 and R=0.7).6. Plan to write a few standard jet selections to ESD (e.g., R=0.7 , R=0.4 cone, Kt)7. Investigate other improvements in jet-finding and jet calibration if time permits.
• improved definition of truth.• improved noise suppression techniques.• more extensive studies of jet-finding with topological clusters.• additional parameters in sample weighting.
20-Dec-04F. Merritt
23
New method of studying truth definition (S. Padhi, F. Paige)
• Select a single jet from dijet production using MC “truth” jets, and propagate only those particles into the detector.
• This separates energy calibration/resolution from difference in clusterization between truth and reconstruction.
• Initial results look promising.
20-Dec-04F. Merritt
24
Kyle Cranmer: Local noise suppression using neighboring cell energy:
20-Dec-04F. Merritt
25
20-Dec-04F. Merritt
26
20-Dec-04F. Merritt
27
Improving sampling wt’s(A. Gupta)
• Using sampling weight for each
calorimeter layer is not very
useful
-- large fluctuation in a single
layer.• But using fraction of
energy deposited in EM and HAD
have useful information on how
jets develops.• To make weights use energy
fraction information in EM and
HAD calorimeter.
25 GeV
100 GeV
400 GeV
1000 GeV
Fraction of Jet energy in EM and HAd
20-Dec-04F. Merritt
28
Energy fraction
LAr/(LAr+Tile) < 0.67
Energy fraction
LAr/(LAr+Tile) > 0.67
20-Dec-04F. Merritt
29
Study variations in calibration for different physics processes (F.P.)