CALOR 2004 - 1-April-20041

Algorithms for the DØ Calorimeter

Sophie Trincaz-DuvoidLPNHE – PARIS VI for the DØ collaboration

Calorimeter short description Algorithms at the cells level Jets reconstruction Electromagnetic objects reconstruction Missing transverse energy

CALOR 2004 - 1-April-20042

The DØ calorimeter

Liquid argon active medium and uranium absorber Hermetic with full coverage :|| < 4.2 ; int > 7.2 (total) Transverse segmentation (towers) x = 0.1x0.1 Energy resolution e : E/E = 23%/E + 20%/E + 0.4% jets : E/E = 90%/E + 5.2%

x

z

y

End Calorimeter (EC)

Central Calorimeter . (CC)

Coarse hadronic (CH)

Fine hadronic (FH)

Electro- magnetic (EM)

55000 cells 50 dead channels

CALOR 2004 - 1-April-20043

Off-line noise reduction : NADANADA : New Anomalous Deposit AlgorithmPurpose : Identify anomalous energy

deposition in the calorimeter : the isolation of high PT cells is checked by looking at their neighbors.

depth(,) plane

PTcell < -1 GeV or PTcell > 500 GeV or 500 GeV > PTcell > 1 GeV and Eneighbors < Ecut

if the cell is rejected

0.04 mis-identified cell per event

distribution of rejected cells shows accumulations

MET (before rejection)

<> :484 GeV

MET (after rejection)

<> :6.4 GeV

Effect of NADA on Missing Transverse Energy (MET)

CALOR 2004 - 1-April-20044

Off-line noise reduction : T42 Purpose : improve the ratio signal to noise by rejecting .. isolated cells of low energy and cells of negative energy. The algorithm T42 rejects : . all isolated cells with energy below 4 of noise . all negative cellsIt keeps : only high signal cells (with energy above 4) . their significant neighbors (between 2.5 and 4) Allow to improve the missing transverse energy and the reconstruction of jets and e/ at low energy

before T42after T42

Number of cells per event

before T42after T42

Cells energy in number of

CALOR 2004 - 1-April-20045

Objects reconstruction

Jets Electromagnetic objects Missing transverse energy

CALOR 2004 - 1-April-20046

Jets reconstruction algorithm Improved Legacy Cone Algorithm Pre-clustering = Seeds for clustering. Simple cone algorithm : A first clusterisation is made around high PT towers (> 500 MeV) in a cone of R=0.3 Tower size = 0.10.1 in (,)

Rejection of fake clusters : If the higher PT cell of a tower is in the CH and if Ecell < Ethreshold then the tower is considered only if PT(tower)-PT(cell) > 500 MeV Clustering : simple cone algorithm (R=0.5 or R=0.7)Seeds are in the center of a R cone which is iteratively drifted until it is stable. PT jet > 8 GeV If cones overlap apply a split/merge algorithm.

splitmerge

CALOR 2004 - 1-April-20047

Jets identificationVariables that help to reject fake jets. PT fraction of jets in the CH : < 0.4 PT fraction of jets in the EM : 0.05 < EMF < 0.95 Number of towers containing 90% of the jet energy : > 1 PT of the higher cell / PT of the second higher cell : < 10 Use the L1 information : around the center of a jet, sum the transverse energy of L1 towers within a cone (PT(L1)). PT(L1)/Ptjet > 0.4 (0.2 in the intercryostat detector region) .

Jet efficiency reconstruction+identification : 95 % 5 % on data for PT>20 GeV 95 % on MC for PT>20 GeV 98 % on MC for PT>25 GeV MC

CALOR 2004 - 1-April-20048

Variables for jets identification

CHF<0.4

Bad jets Good jets

0.05<EMF<0.95

L1 confirmation

Efficiency for fake jets after L1 confirmation

old jet-Id

new jet-Id

Good jets sample: di-jet eventFake jets sample: events with . more than 5 jets

PT fraction

of jets in the CH (CHF)

PT fraction of jets in the EM (EMF)

Effect of L1 confirmation

CALOR 2004 - 1-April-20049

Jets in DØ RunII analysis

GeV

Data – MC comparison for jets PT in W+dijets events used for Wbb studiescone < 0.5

• data MC

Dijet mass cross section data/theory comparisoncone < 0.7

Inclusive jet PT cross section data/theory comparisoncone < 0.7

More informations

CALOR 2004 - 1-April-200410

Reconstruction of high PT electronsThe algorithm is simple cone algorithm : Electromagnetic clusters are obtained by adding towers within a cone of R<0.2 around seeds ( towers of higher transverse energy (PT>0.5 GeV)). Cluster criteria is : PT(cluster)>1.5 GeV and EMF>0.9. The estimators used to identify electrons are : • Isolation of the cluster : Energy between cones R<0.2 and R<0.4 must be low : E<E*0.15• The shower shape probability is evaluated on data (computing 7 variables) and compared to the simulation. • Track matching (2 probability must be reasonable)Cut on a Likelihood using 8 variables to identify electrons. The variables include the 3 estimators and ET/PT, the DCA, the number of tracks in a cone R<0.05, the total PT of tracks in a cone R<0.4 around a candidate track and the number of CPS strip hits per electromagnetic energy. Electron reconstruction efficiency : (84.10.9) % on data in CC . (91.5 1.3) % on data in EC

CALOR 2004 - 1-April-200411

Electrons reconstruction

Z e+e-

• Data

MC

Z mass resolutionData : 3.8 % in CC – 3.1 % in EC MC :2.3 % in CC – 1.6 % in EC

Probability for a fake electron to have a track matched to it as function of PT

Central Calorimeter

End Cap

CALOR 2004 - 1-April-200412

Reconstruction of low PT electronsThe road method :Developped for non-isolated and low PT electrons Each reconstructed track of PT>2 GeV is extrapolated in the calorimeter

The road is reconstructed with cells in a cylinder around the track extrapolation (use the calorimeter granularity). Discriminating variables : • Electromagnetic fraction : EMF > 0.85• E/p : 0.6 < ET(3 first EM layers)/PT(track) <1.05Efficiency : ~82 % on MC for PT>5GeV

J/ e+e-

CALOR 2004 - 1-April-200413

Photons reconstructionPhotons are reconstructed with same method as high PT electrons (simple cone algorithm). They differ from electrons by the absence of a track. So currently photon identification is done by requiring electrons without track matching. Tighter criteria are adjusted by analysis groups.

Transverse energy of photons for studies of W production.

Example of studies of W production. Criteria for photons identification are : PT (tracks in the cone R<0.4) < 3 GeVPhotons are reconstructed in the central calorimeterPT(photon) > 8 GeV

CALOR 2004 - 1-April-200414

Reconstruction of Missing Transverse Energy

Purpose : identify a particle which doesn’t interact in the detector (such as )Quadratic sum of cells transverse energy Visible Transverse Energy (VET)Transverse Energy = 0 Missing Transverse Energy (MET) = -VET

VET is computed with all cells excepted those from CH The CH part of jets is added to VET The bad jets are removed from VET VET is corrected from electromagnetic scale and jet energy scale Muons transverse energy is removed from MET.

CALOR 2004 - 1-April-200415

Missing Transverse Energy

Missing transverse energy in events Z. Data/simulation comparison

• Data MC MET

METyMETx

Missing transverse energy in W+dijets events used for Wbb studies

CALOR 2004 - 1-April-200416

Conclusion

For the jets, standard cone algorithm is used with 0.5 and 0.7Reduction of main systematics will come from improved jet energy scale. Electrons reconstruction/identification well understood for high PT physicsFurther efforts on low energy and non-isolated electrons Use pre-shower detector information to improve photon identification Improved missing transverse energy from off-line noise treatment Improvements are still to come and great results are in front of us.

CALOR 2004 - 1-April-200417

Backup slides

CALOR 2004 - 1-April-200418

Effect of the T42 algoritm on jets

Jet multiplicity lowered by 0.5 jets per event

before T42after T42

Number of jets per event

Sample of events used for top reconstruction in all-jets channel

before T42after T42

Number of jets merging per event

PT of jetsbefore T42after T42