500 likes | 635 Views
CBM Calorimeter System. CBM collaboration meeting, March 2009 I.Korolko (ITEP, Moscow). Outline. ■ Reconstruction in the CBM ECAL M.Prokudin ■ Reconstruction of π 0 and η mesons S.Kiselev ■ Two photon reconstruction and low mass background K.Mikhailov (A.Stavinsky).
E N D
CBM Calorimeter System CBM collaboration meeting, March 2009 I.Korolko (ITEP, Moscow)
Outline ■ Reconstruction in the CBM ECAL M.Prokudin ■ Reconstruction of π0 and η mesons S.Kiselev ■ Two photon reconstruction and low mass background K.Mikhailov (A.Stavinsky)
Our main efforts during last year 1) Development of reconstruction for the CBM ECAL • Need it for optimization and physics feasibility studies • Quality of reconstruction (at high multiplicities) • “Popularization” of ECAL reconstruction 2) Optimization of the CBM ECAL (reducing price)
“Optimized” calorimeter Main features: • ~14K channels • Efficient γ, π0, η reconstruction • Electron identification • Movable design (no central region)
Reconstruction • Requirements- fast and robust at CBM multiplicities • Cluster finder algorithm and performance • Cluster fitting performance • Matching with tracks and photons for debugging and efficiencies
Reconstruction. Cluster finder Cluster formation • Remove maximums near charged tracks • Use real tracking • Precluster: • formed near local maximum • cut on maximum energy • find maximum 2x2 matrix near maximum • add a neighbor to local maximum cell with minimal energy deposition • to add information • check precluster energy • >0.5GeV • Cluster: group of preclusters with common cells ------------------------------------------------- • Clusters with 2 maximums >10% • Fit procedure is necessarily! • >3 maximums can be omitted Requirements • Clusters should be large • information for unfolding • Clusters should be small • hadrons background
Reconstruction. Fitter • Isolated photons – most simple case • Two close photons: • Robust reconstruction in case of two separate maxima • Recognition of two photons in case of one maximum (χ2 criteria) • χ2 shape should not depend on photon’s energy • same efficiency value for the chosen cut • separation power of one/two photons in case of one maximum is an important criteria • example: with 95% efficiency for clusters formed by single photon
Reconstruction. Fitter basics Same photons, but different distance between them. 2 separated clusters • Trivial case cluster with 2 maxima • Shower shape fit Cluster with 1 maximum (2 γ) • Should be rejected • Shower shape fit
Reconstruction. Shower shape + or ? Precise knowledge of shower shape is essential. If χ2 is used, than what is the error σ2 ?
Reconstruction. Shower library Cell 4 • Store mean energy deposition in small cells vs. (x, y) for each • Energy • Theta • Phi • Energy depositions in cluster cells are not independent • RMS value storing is useless • Trying analytical formula for σ2 • Hoping to take into account correlations Cell 5
Reconstruction. Errors σ2=c2(Emeas(1-Emeas/Ecluster)+c0+c1E2cluster) • c2 is normalization • 95% of photons have χ2<2 • c1 and c0 are determined requiring that χ2 • does not depend on photon’s energy • Does not depend on φ angle • ci is different for each calorimeter region
Reconstruction. Performance Single (isolated) 1 GeV photons 7.3% Calibration is perfect (see later…)
Reconstruction. Performance Two close 1 GeV photons (forming 2 maxima) • Reconstructed energy distribution for both photons is Ok • But for each of them… 8.9%
Reconstruction. Performance Two close 1 GeV photons (forming 2 maxima) Reconstruction algorithm tends to increase asymmetry in energy of photons (also by PHENIX experience)
Reconstruction. 2 γ recognition Inner calorimeter region • χ2 criteria allows identify ~40% of two photon clusters • efficiency highly depend on distance between photons
Reconstruction. Performance AuAu 25 GeV UrQMD events • 730 photons in event • 299 photons in calorimeter acceptance • 131 with energy > 0.7 GeV • 35% reconstruction efficiency • 91% for isolated tracks • rises with increasing isolation
Reconstruction. Performance AuAu 25 GeV UrQMD events • 35% reconstruction efficiency • Boundaries between calorimeter regions • occupancy Reconstruction efficiency vs. θand energy
Reconstruction. Matching 1 • Idea: use shape of reconstructed particles • For each MC/reconstructed particle compute: • only cells with energy deposition from current particle are in play • match with MC particle with Pi>0.6 • Clusters with 2 maximums: 99.97% efficiency
Reconstruction. Matching 2 • Idea: for γ (e±) look at mother e± (γ) and grandmother and … • for each γ and e± MCTrack: P=Pthis+ΣPdaughter • Several realizations available • Choose γ/e± with maximum P • Choose parent e± if daughter γhave P>const*Pγ • … • Exact algorithm/constants are defined in configuration file • still under development γ e- e+ γReco
Reconstruction. Conclusions • Reconstruction algorithms are completed and tested • 35% reconstruction efficiency • occupancy! • Calorimeter geometry optimization • cost • physical observables sensitivity • reconstruction efficiency • Digitization and response nonuniformity impact • moving towards realistic geometry
Reconstruction of π0 and η mesons • Sergey Kiselev, ITEP Moscow for the ECAL group • Input info • Spectra and acceptances • Ideal reconstruction • Real reconstruction • efficiency • true signal, S/B • extracting signal by mixing • Summary
π0 and η mesons. Input info • CbmRoot package (trunk JAN09), Geant3 • 2 104UrQMD Au+Au central events at 25 AGeV • simulated and reconstructed in ECAL by Misha Prokudin • the ECAL wall at 12 m from a target • Size: X x Y = 12 x 9.6 m2 , beam hole 0.8 x 0.8 m2 • pγ cut: pγ > 0.3 GeV/c • Cluster cut: χ2 < 3
π0 and η mesons. Vertex γ spectra acceptance (%) <accep.> = 50%
π0 and η mesons. Primary π0 spectra acceptance (%) <accep.> = 12%
π0 and η mesons. Primary η spectra acceptance (%) <accep.> = 9%
π0 and η mesons. Vertex γ 386 reco γ/event: 30 not matched with MC tracks 356 matched with MC tracks 237 of them Rvtx <0.1 cm 160 of them are photons 98 of them enter ECAL 62 “enter” ECAL by decay products 1738 MC tracks/event enter ECAL 398 of them are photons 230 of them Rvtx <0.1 cm “vertex” γreco efficiency = 98 / 230 = 43%
π0 and η mesons. Vertex γ reconstruction efficiency peaks at θ=70 and 120 because of change in the cell size
π0 and η mesons. Primary π0 reconstruction efficiency 20.1 reco primary π0/ev.: 6.3: 2γ enter ECAL 7.9: 1γ enter ECAL 5.9: 0γ enter ECAL 364 primary π0 /ev.: 32.7 enter ECAL primary π0reco efficiency = 6.3/32.7 = 19 (%)
π0 and η mesons. Primary η reconstruction efficiency 1.56 reco primary η: 0.43: 2γ enter ECAL 0.62: 1γ enter ECAL 0.51: 0γ enter ECAL 14.3 primary η 2γ : 1.6 enter ECAL primary π0reco efficiency = 0.43/1.6 = 27 (%)
true S from primary π0 2γ/ECAL: higher Mγγ 1γ+0γ/ECAL: lower Mγγ sum: rather Breit-Wigner than Gauss fit real reco: ~5 times broader signal, 37/30/25/23 MeV, than for ideal reco
true S from primary η The same remarks as for π0 real reco: ~5 times broader signal, 100/102/90/63 MeV, than for ideal reco
ideal vs real S/B S/B2σ(%) (signif.)
π0 and η mesons from another analysis Still some good luck and fine tuning are required…
extracted (S+B)-Bmix 5 mixed events to evaluate Bmix (S+B) and Bmix were normalized at M>0.3 GeV at high pt (S+B)-Bmix and true S are in reasonable agreement For η higher statistics needed
Summary 2 104 UrQMD Au+Au central events at 25 AGeV with ECAL reco high pt π0 can be extracted by mixing event technique η: ~ 2 order higher statistics needed Recommendation: test reconstruction at lower system/energy
Two photon reconstruction with ECAL and low mass background Alexei Stavinskiy, Konstantin Mikhaylov ITEP, Russia
Input 2*104 central UrQMD events AuAu@25AGeV (Local analysis) Full ECAL reconstruction (version of February 2009) with CBM root January 2009 version (version with new geometry) Cuts: Minimal energy deposited in ECAL 500 MeV. χ2 (of photon reconstruction) < 3. Minimal distance between cluster (DBC) > 20 cm. Particles from target = Vz < 1cm (conventional)
WA 98 experiment [arXiv:nucl-ex/0006007] A low mass tail on the π0 was observed. The tail can result from π0 produced downstream from target: • from decays (K0s, ...) • from background interaction on downstream materials (the normalized target out background contribution is shown by the open circles In part c))
1t 1b θb θt 2b πtarget πbackground 2t Differences between target and background pairs: *mean photon energy higher for target photons *real emission angle difference for target photons from π(η) decay corresponds to measured invariant mass; this is not the case for background pairs *vertex position is fixed for target pairs; for background pairs vertex position distribution corresponds to detectors (support) position
Mγγ vs Vz π0 ECAL RPC TRD3 TRD2 TRD1 RICH
Ecut Rcut: Mγγ,same mother S/B for 0<Mγγ<120MeV
DBC cut: Mγγ,same mother Distance Between Clusters>20cm Distance Between Clusters>50cm
Signal-Background VZ<1cm S/sqrt(S+B)=20 S/B=0.25%
Two-photon correlations e+ 2 2rec 1 1 e- External conversion: • No close cluster interference • No hadron contamination C2 calculated in EMCAL and converted+EMCAL agree => both effects are under control
Measurement of Direct Photons via Conversion in CBM Melanie Klein-Bosing WWU Munster,Germany CBM Collaboration Meeting, Dubna 2008, October
Conclusions • The feasibility of π and η meson identification with ECAL was shown • The low mass background in two photon invariant mass was studied • Two main contribution to the low mass background are: • Interaction with downstream detector construction • Decay of long lived particles • Possible cuts to reduce (slightly) low mass background: • Cut on gamma energy • Cut on distance between center of clusters • Other possible ways: • Background simulations • Combined photon pair identification with different detectors
Conclusions 1) Reconstruction in ECAL is 90% ready • better definition of errors • development of matching algorithms (converted γ) • number of users = quality 2) Use fast (25 ns) ECAL for triggering (Alla)