e/e/ Calibration and Calibration and Performance with Performance with
DC1DC1
Y.Q.Fang, K.Loureiro, B.Mellado, Y.Q.Fang, K.Loureiro, B.Mellado, S.PaganisS.Paganis, W.Quayle, Sau Lan Wu, W.Quayle, Sau Lan Wu
Univ. Of Wisconsin, MadisonUniv. Of Wisconsin, Madison
29/6/04 2
IntroductionIntroduction
A big part of our group is involved in Higgs A big part of our group is involved in Higgs analysesanalyses
Thus we have large DC1 samples with W,Z Thus we have large DC1 samples with W,Z e/g jets etc (e/g jets etc (all available in castorall available in castor))
When attempting physics analyses with full When attempting physics analyses with full sim, one has to deal with calibration issues sim, one has to deal with calibration issues which is obvious for e/which is obvious for e/ p pttMiss etc.Miss etc.
So, typically one stops and starts studying So, typically one stops and starts studying calibration issues, looking at TBeam results.calibration issues, looking at TBeam results.
This may be a “top->bottom” approach but This may be a “top->bottom” approach but several people in ATLAS are trying to do several people in ATLAS are trying to do physics analyses with full-sim: we decided to physics analyses with full-sim: we decided to check ideas obtained by EMBarred TBeam check ideas obtained by EMBarred TBeam studies and work closer with some LAr studies and work closer with some LAr experts.experts.
29/6/04 3
ATLAS TB-MC: T.Carli offset makes possible simultaneous optimization of linearity and resolution
:E E :Set truecalocalo
0.005/E
:timprovemen
Possibility to get • good resolution• good linearityEffect bigger thanin Toy-MC !?
linearity optimised resolution optimised
29/6/04 4
Electrons (DC1)Electrons (DC1)
33210 EWEEEWbE presrec Extract Extract ,b,W0 and W3 in 55 ,b,W0 and W3 in 55 bins bins <0.025<0.025 Apply them at the CBNT sampling energiesApply them at the CBNT sampling energies Make sure everything done in ATHENA is Make sure everything done in ATHENA is
correctedcorrected
29/6/04 5
For DC1 we obtained new sets For DC1 we obtained new sets of electron-based EM weightsof electron-based EM weights
What was done and how What was done and how including the extracted including the extracted weights is described inweights is described in
ATL-COM-GEN-2004-003ATL-COM-GEN-2004-003 There were requests by There were requests by
people who do analysis for people who do analysis for methods doing these methods doing these corrections at CBNT level.corrections at CBNT level.
SP provided the code and he SP provided the code and he can help anybody who wants can help anybody who wants to try this out in his analysisto try this out in his analysis
DC2: DC2: NO PROBLEMNO PROBLEM cluster cluster level corrections can be level corrections can be trivially applied (they are trivially applied (they are not hardcoded)not hardcoded)
29/6/04 6
Lower energy electronsLower energy electrons
Linearity is improved without Linearity is improved without deterioration of the resolutiondeterioration of the resolution
29/6/04 7
High and very high energy High and very high energy electronselectrons
• Thanks to Martina Schaefer (Grenoble) who applied the weights to an analysis which uses high energy electrons (here the offset plays no role)
• Application to Z’->e+e- (exotics group analysis)
• Good resolution and linearity is crucial for fitting the mass and discriminating between different models
29/6/04Higgs meeting 12 may 2004 Martina Schäfer 8
Calibration (1)Calibration (1) “standard” calibration :
photons de-calibration re-calibration only barrel
Stathes Paganis
before
recalibration
after
recalibration
/E=0.9%
(E)/E (E=200GeV) =9.5%sqrt(E)-1 + 0.45% 0.8% ok
energy
200GeV
29/6/04Higgs meeting 12 may 2004 Martina Schäfer 9
Calibration (2)Calibration (2)
before
recalibration
after
recalibration
/E=0.8%
(E)/E (E=1000GeV) =9.5%sqrt(E)-1 + 0.45% 0.5%
energy
1TeV
29/6/04Higgs meeting 12 may 2004 Martina Schäfer 10
Calibration (3)Calibration (3)Results on the Z’ (SSM 1.5TeV), electrons at about 750GeV
(E)/E (E=750GeV) =9.5%sqrt(E)-1 + 0.45% 0.6% ok
(M)/M (M=1.5TeV) = sqrt(2) (E)/E 0.8% ok
resolution on the mass (for 1.5TeV)
= 11 GeV
+ tails
/E 7.5%
/E0.7%
resolution of electrons (Z’ at
1.5TeV)
29/6/04Higgs meeting 12 may 2004 Martina Schäfer 11
Total decay width (3)Total decay width (3)1.5TeV
/GeV
all models (generation)
full sim, SSM 1.5TeV
Mll/GeV
DY
M recalibrated
fit
29/6/04 12
Conclusion on electrons: Conclusion on electrons:
(Obvious): calibration has to be e-(Obvious): calibration has to be e-basedbased
At low energies the “offset” parameter At low energies the “offset” parameter improves the linearityimproves the linearity
The extracted weights with lower The extracted weights with lower energy electrons seem well behaved at energy electrons seem well behaved at high and very high energieshigh and very high energies
Impact on Physics: certain analyses Impact on Physics: certain analyses (like the Z’(TeV)) cannot be done with (like the Z’(TeV)) cannot be done with full-sim without proper calibrationfull-sim without proper calibration
29/6/04 13
Photons: can anything be Photons: can anything be done? (DC1)done? (DC1)
29/6/04 14
e/e/ dependent Calibration Algorithm dependent Calibration Algorithm(a first simple criterion)(a first simple criterion)
33210 EWEEEWbE presrec
33210 EWEEEWE presrec
For track-matched clusters we apply electron based calibration:
For un-matched clusters we apply photon based calibration:
29/6/04 15
Correction un-matched EM clusters Correction un-matched EM clusters (XKalman tracks)(XKalman tracks)
DC1 Calibration After Correction(Photon specific)
29/6/04 16
Calibration for converted Photons using Calibration for converted Photons using track-match tag track-match tag
Before Calibration After Calibration(electron calibration)
Tail? dominated by early asymmetric conversions?
29/6/04 17
Early asymmetric Vs relatively late Early asymmetric Vs relatively late symmetric conversionssymmetric conversions
Relatively Late: Rv>150mm
Symmetric: 7.0)/( eeeEEE
3.0)/( eeeEEEand
Very Early: Rv<150mm
Asymmetric: 8.0)/( eeeEEE
2.0)/( eeeEEEor
29/6/04 18
Calibration Performance: photon Calibration Performance: photon linearity linearity
Clear improvement in Linearity is seen
29/6/04 19
Conclusion on photons: Conclusion on photons:
It is encouraging that a pid specific It is encouraging that a pid specific algorithm may improve linearity/resolutionalgorithm may improve linearity/resolution
But, we don’t really understand what are But, we don’t really understand what are the underlying effects:the underlying effects:
Most (80%) of the photons convert, mostly asymmetrically
But only a small fraction of them (10%?) decay early enough to be identified by the ID.
So in ATLAS the photon is really ~2 electrons! We also have ~5% non-interacting photons (no PS
signal) Topological clustering should provide Topological clustering should provide
more info (Boonekamp et al)more info (Boonekamp et al) TBeam data important (but systematic TBeam data important (but systematic
errors may be large)errors may be large)
29/6/04 20
DC2: what’s next + work-DC2: what’s next + work-planplan Need electron Samples out to |Need electron Samples out to ||=2.5|=2.5
We produced 20,50,100 GeV for ||<1.37 (in Castor)
There will be some official major production soon
We need a ClusterCorrectionToolWe need a ClusterCorrectionTool This should read from a file or dBase and apply
corrections at the cluster level (Karina Loureiro+...)
Look more carefully at photons and Look more carefully at photons and conversions (using e/conversions (using e/ samples) samples)
TestBeam04 cross-checks (TestBeam04 cross-checks (KL+SPKL+SP)) Impact on Physics (difficult without Impact on Physics (difficult without
full validated samples)full validated samples)