Status of Status of calorimeter calorimeter simulationssimulations

Mikhail Prokudin, ITEPMikhail Prokudin, ITEP

OutlineOutline

►ππ˚̊ reconstruction quality reconstruction quality►Cluster findingCluster finding

requirements requirements 2 approaches2 approaches

►Shower libraryShower library►Simplified MC supporting design studiesSimplified MC supporting design studies

by A. Chernogorov

►Next stepsNext steps

ππ˚̊ reconstruction quality reconstruction quality

► Standard 25GeV Standard 25GeV AuAu UrQMD AuAu UrQMD eventevent

► γγ from same from same ππ˚̊► Smooth withSmooth with

7%/sqrt(E) 7%/sqrt(E) energy resolutionenergy resolution

3cm position 3cm position resolutionresolution

both both

RMS 2.627e-3

RMS 6.669e-3

RMS 6.692e-3

3cm position resolution doesn’t affect the reconstruction quality!

Requirements for cluster finding Requirements for cluster finding procedureprocedure

► Fitting routineFitting routine first approximationfirst approximation

► energyenergy calibrationcalibration

► positionposition S-curvesS-curves

► number of photons number of photons number of local number of local

maximums in clustermaximums in cluster

iterative procedureiterative procedure► shower shapeshower shape

shower libraryshower library► shower widthshower width

► Cluster findingCluster finding maximize cluster sizemaximize cluster size

► use all possible use all possible information for fittinginformation for fitting

► minimum 3 cells for minimum 3 cells for cluster with 1 cluster with 1 maximummaximum

3 degrees of freedom 3 degrees of freedom per photonper photon

occupancyoccupancy► hadron contaminationhadron contamination

minimize number of minimize number of maximumsmaximums

► less parameters in fitless parameters in fit stabilitystability

Cluster unfolding. Naive Cluster unfolding. Naive approachapproach► Color: energy Color: energy

depositiondeposition► Edging:Edging:

Green: clustersGreen: clusters Red: excluded by Red: excluded by

charged trackscharged tracks► Cluster Cluster

definition:definition: connected area connected area

of cells with of cells with energy energy deposition above deposition above the thresholdthe threshold

► Threshold Threshold ~150MeV~150MeV

BTW, Real tracking!

Cluster finding. Naive Cluster finding. Naive approachapproach

► Cluster definition:Cluster definition: connected area of cells with connected area of cells with

energy deposition above the energy deposition above the thresholdthreshold

threshold can be found from threshold can be found from MCMC

► Occupancy differs ~10Occupancy differs ~1022 times times► partially compensated by partially compensated by

segmentationsegmentation lost clusters in periphery of lost clusters in periphery of

calorimetercalorimeter create too large clusters in create too large clusters in

central regioncentral region► No chance to fit cluster with No chance to fit cluster with

>>77 peaks peaks central region is lost!central region is lost!

► Clusters with size <3 cells Clusters with size <3 cells require additional require additional informationinformation

ALICE approach

25 UrQMD events

Peaks per cluster

Cluster findingCluster finding

► Find peak above thresholdFind peak above threshold peaks associated with charged track are peaks associated with charged track are

removed from the considerationremoved from the consideration► Construct an area with certain number of Construct an area with certain number of

cells around the peakcells around the peak it should contain 2x2 submatrix with maximum it should contain 2x2 submatrix with maximum

energy of 3x3 matrix around peakenergy of 3x3 matrix around peak size determined bysize determined by

► reconstruction quality of single photonreconstruction quality of single photon at least 3 cells for reconstructionat least 3 cells for reconstruction at least 4 cells for unfoldingat least 4 cells for unfolding

► occupancyoccupancy► Find areas intersecting with given areaFind areas intersecting with given area► Construct a clusterConstruct a cluster

PHENIX-like approach

Cluster unfoldingCluster unfolding

► Color: energy Color: energy depositiondeposition

► Edging:Edging: Green: clustersGreen: clusters Red: excluded Red: excluded

by charged by charged trackstracks

► Less clusters in Less clusters in central region central region with new with new algorithmalgorithm

► Cluster size Cluster size under controlunder control

ComparisonComparison

Naive New25 UrQMD events

Number of peaks per cluster lower using new approach

Peaks per cluster

Conclusions for clustering Conclusions for clustering algorithmalgorithm

►Naive approach does not workNaive approach does not work occupancy variationsoccupancy variations

► New (PHENIX-like) approach New (PHENIX-like) approach number of peaks in cluster under controlnumber of peaks in cluster under control less sensitive to occupancyless sensitive to occupancy

►Real trackingReal tracking

Shower shapeShower shape

► Analytical formula for shower shape approximationAnalytical formula for shower shape approximation ALICE and PHENIX experienceALICE and PHENIX experience No memory consumptionNo memory consumption

► Best for multicore CPUBest for multicore CPU Poor quality for large incident Poor quality for large incident

anglesangles► Shower libraryShower library

Fits exactly to the dataFits exactly to the data Requires a lot of memoryRequires a lot of memory

► (and CPU!)(and CPU!) Any incident angleAny incident angle checking analytical approximation qualitychecking analytical approximation quality

► Shower widthShower width Also needed for fitting procedureAlso needed for fitting procedure Approximation or storing in library?Approximation or storing in library?

Shower libraryShower library

► ECAL with very high ECAL with very high segmentation (1x1mmsegmentation (1x1mm22 volumes) volumes) use one shower multiple timesuse one shower multiple times volumes merging procedurevolumes merging procedure

► Transport photons for every:Transport photons for every: EnergyEnergy ThetaTheta PhiPhi

► Eightfold decrease due to symmetryEightfold decrease due to symmetry► For low energies we have to For low energies we have to

generate more showersgenerate more showers► Larger fluctuations for low energy Larger fluctuations for low energy

showersshowers 10k for 0.49 GeV10k for 0.49 GeV 2k for 16GeV2k for 16GeV

Volumes mergingVolumes merging

► During shower During shower library creationlibrary creation Faster during Faster during

reconstructionreconstruction Also can store RMS Also can store RMS

(see next slide)(see next slide) Bigger library sizeBigger library size Cell size fixedCell size fixed

► Different data set for Different data set for each cell sizeeach cell size

► During During reconstructionreconstruction Homogenous datasetHomogenous dataset

► Cell size not fixedCell size not fixed

Information about Information about RMS is lostRMS is lost

► Need analytical Need analytical approximation for approximation for errorserrors

More CPU required More CPU required during reconstructionduring reconstruction

Shower widthShower width

► Energy deposition in Energy deposition in cluster cells are not cluster cells are not independentindependent RMS value storing RMS value storing

uselessuseless

► Need an analytical Need an analytical formulaformula correlations should be correlations should be

in!in!

► ……

h4 h5

h4

h5

Shower libraryShower library

► 170 Mb disk space170 Mb disk space► 300 Mb in 300 Mb in

memorymemory► Energy: 0.49, 1, 2, Energy: 0.49, 1, 2,

4, 6, 9, 12, 16 4, 6, 9, 12, 16 GeVGeV

► Theta: 2Theta: 2˚̊, 3.5, 3.5˚̊, , 55˚̊, 6.5, 6.5˚̊, 8, 8˚̊, 9.5, 9.5˚̊, , 1111˚̊, 13, 13˚̊, 16, 16˚̊, 20, 20˚̊, , 2424˚̊, 28, 28˚̊, 32, 32˚̊

► Phi: 0Phi: 0˚̊, 10, 10˚ ˚ , 20, 20˚̊, , 3030˚̊, 40, 40˚̊

See CbmShowerLib class

6x6cm6x6cm22 cells cells

Shower libraryShower library12x12cm12x12cm22 cellscells

► Shower rotating on fly?Shower rotating on fly? classical trade CPU vs. memoryclassical trade CPU vs. memory

Simplified MC supporting design Simplified MC supporting design studiesstudies

► Constructive detailsConstructive details steel bandsteel band dead material in dead material in

frontfront fibers insertionfibers insertion

► Necessarily for Necessarily for design worksdesign works can be used in can be used in

CBMROOT simulationCBMROOT simulation

Ex. 1. SS bands between modules on the ECAL response. Eγ=16 GeV, θ=3° Pb=Sc=1mm . 1- ▲, 2-*, 3-●

Steel band effects

by A. Chernogorov

Dead material in front of the Dead material in front of the calorimetercalorimeter

► CalorimeterCalorimeter 200 layers200 layers 1 mm Pb + 1 mm Sc1 mm Pb + 1 mm Sc

► AbsorberAbsorber C, Fe, and Cu studiedC, Fe, and Cu studied 0.0 X0.0 X00, 0.2 X, 0.2 X00, 0.4 X, 0.4 X00, 0.8 , 0.8

XX00

distance to ECAL 2-10mdistance to ECAL 2-10m

by A. Chernogorov

ECAL5m

γ

Example: Fe, 5m

Conclusions and next stepsConclusions and next steps

► Real tracking in calorimeterReal tracking in calorimeter► Studied different approaches to cluster findingStudied different approaches to cluster finding

naive one does not worknaive one does not work► Shower library is completeShower library is complete

realization in SVN and AUG07realization in SVN and AUG07► Unfolding procedure under (extensive) Unfolding procedure under (extensive)

developmentdevelopment analytical formula for shower widthanalytical formula for shower width

► with correlationswith correlations can be submitted to SVN any momentcan be submitted to SVN any moment

► Design supporting MC studies under wayDesign supporting MC studies under way by A.Chernogorovby A.Chernogorov

Procedure of Procedure of γγ reconstruction reconstruction► First approximationFirst approximation

energyenergy► calibrationcalibration

positionposition► S-curvesS-curves

► Cluster unfoldingCluster unfolding cluster finding cluster finding

procedureprocedure fitting routinefitting routine

► shower shapeshower shape shower libraryshower library

► LHCb like methodsLHCb like methods► Pure Pure γγ,, no background no background► Simple and easy to checkSimple and easy to check► Test site for shower library Test site for shower library

routinesroutines► Can be done in few monthCan be done in few month

► ALICE PHENIX-like methodsALICE PHENIX-like methods► Require much more effortRequire much more effort

CALO parameters should be fixed?CALO parameters should be fixed?

From September 2006 CBM collaboration meeting

Don

e

From September 2006 CBM collaboration meeting