Application of the HBOM method
Transkrypt
Application of the HBOM method
Correcting for detector effects in high-multiplicity events with the HBOM method Piotr Janus, Jakub Kremer AGH University of Science and Technology in Kraków Kraków, 8.1.2016 Detector effects Measurements of particle multiplicities in high energy collisions may be affected by following detector effects: inefficiency in the detection of particles, incorrectly reconstructed tracks (called fakes), secondary particles created in decays of primary particles or due to their interactions with the detector material. In general measurements with a detector can be expressed by: Z g (y ) = A(y , x)f (x)dx, where: g (y ) - measured distribution f (x) - truth distribution A(y , x) - response function Discrete form with n bins for x and m bins for y : y = Ax y - measured histogram x - truth histogram A - response matrix Piotr Janus, Jakub Kremer (AGH) Application of the HBOM method Kraków, 8.1.2016 2 / 11 General idea of unfolding 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 -1 Response matrix truth arb. unit To correct for detector effects an unfolding method is used which needs to invert the response matrix: x = A−1 y truth measured PDF 5 4 3 2 1 0 0 1 2 3 -1-1 4 5 observable 0 1 2 3 4 5 measured 14 12 10 8 6 4 2 0 arb. unit However unfolding does not work when real data has a wider distribution than simulation used to obtain the response matrix. Piotr Janus, Jakub Kremer (AGH) 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0-1 mc data 0 1 2 3 4 5 6 7 observable Application of the HBOM method Kraków, 8.1.2016 3 / 11 HBOM (Hit Backspace Once More) method First proposed: J. Monk, C. Oropeza-Barrera, Nucl. Instrum. Meth. A 701 (2013) 17-24 Let us consider a detector response described by operator A, then: X obs = X (0) = AX true = AX (−1) Operator A can be applied i times: X (i) = Ai+1 X true X(i) Repetition of the above procedure will result in a sequence of values of the observable X (i). 5 4.5 4 3.5 3 2.5 2 1.5 1 -1 0 1 2 3 4 5 i Piotr Janus, Jakub Kremer (AGH) Application of the HBOM method Kraków, 8.1.2016 4 / 11 Observables To study performance of the HBOM method factorial moments will be used which are defined as: + * B Y nj ! Fi1 ,...,iB ≡ , (nj − ij )! j=1 where B denotes number of pseudorapidity bins and nj is the number of charged particles emitted into a given pseudorapidity interval. For three pseudorapidity bins, the above equation can be expressed as: FiL ,iC ,iR ≡ nL ! nC ! nR ! (nL − iL )! (nC − iC )! (nR − iR )! In the following we will focus on the four factorial moments: F100 , F101 , F120 , F111 Factorial moments are extracted in 3 pseudorapidity bins with varying distance between forward and backward bin. -2.5 -2 Piotr Janus, Jakub Kremer (AGH) -1.5 -1 -0.5 0 0.5 1 Application of the HBOM method 1.5 2 2.5 η bins Kraków, 8.1.2016 5 / 11 Correction of the factorial moments I To correct for track reconstruction inefficiency the following procedure is applied: measurement of an observable (in this case: factorial moments) using all reconstructed tracks satisfying the analysis cuts – 0th iteration, removal of tracks from the sample according to the detector inefficiencies (discussed in the next slide), repeated measurement of the observable – 1st iteration, removal of further tracks from the sample and measurement of the observable – 2nd iteration, etc., fitting (with exponential or polynomial) the distribution of the observable as a function of the iteration number, extrapolation to the iteration -1 should give an estimate of the true value of the observable. Secondaries and fakes can be also corrected for. Piotr Janus, Jakub Kremer (AGH) Application of the HBOM method Kraków, 8.1.2016 6 / 11 Correction of the factorial moments II counts Condition for removing tracks takes the form: eff > r , 600 where eff is the efficiency for a given track and r is a random number from the uniform distribution in the interval [0,1]. Non-integer iterations can be introduced. In general, the rejecting procedure of the n-th iteration can be done from the 0th iteration with the following condition: 500 eff n > r 10 9 8 7 6 5 4 3 2 1 eff (pT , η) = 2 −1 tg π pT − 80 MeV 150 MeV Piotr Janus, Jakub Kremer (AGH) exp − 0-10% p > 500 MeV T 400 |η|<2.5 300 200 100 2 η 36 Application of the HBOM method 2500 3000 3500 4000 4500 track multiplicity 0.9 0.8 efficiency T p [GeV] 0 For performance studies PbPb collisions at √ sNN = 2.76 TeV were generated using the HIJING generator. As a track reconstruction efficiency function was taken: HIJING PbPb sNN = 2.76 TeV 700 0.7 0.6 0.5 -2.5 -2 -1.5 -1-0.5 0 0.5 1 1.5 2 2.5 η Kraków, 8.1.2016 7 / 11 Closure test I 220 200 180 160 140 120 100 80 60 40 20 0 F101 F100 To correct factorial moments for detector inefficiencies the HBOM method was used. For the fitting function ax 3 + bx 2 + cx + d was chosen. HIJING PbPb s=2.76 TeV Iterations Extrapolation Truth MC HIJING PbPb s=2.76 TeV Iterations Extrapolation Truth MC 25000 20000 15000 10000 5000 |η|<0.1 -0.3<η<-0.1 -1 -0.5 0 0.5 1 1.5 2 2.5 0 3 |η|<0.1 -0.3<η<-0.1 -1 -0.5 0 0.5 1 1.5 2 ×103 3500 HIJING PbPb s=2.76 TeV Iterations Extrapolation Truth MC 3000 2500 2000 0 HIJING PbPb s=2.76 TeV Iterations Extrapolation Truth MC 3000 2000 1500 1000 500 3 ×103 3500 2500 1500 2.5 Iteration F120 F111 Iteration 1000 |η|<0.1 -0.3<η<-0.1 -1 -0.5 0 0.5 500 1 1.5 2 2.5 3 0 |η|<0.1 -0.3<η<-0.1 -1 -0.5 Iteration 0 0.5 1 1.5 2 2.5 3 Iteration In the central rapidity bins the HBOM method reproduces truth value very well. Piotr Janus, Jakub Kremer (AGH) Application of the HBOM method Kraków, 8.1.2016 8 / 11 Closure test II F101 F100 Factorial moments as a function of distance between bin centers corrected with ax 3 + bx 2 + cx + d d. 20000 140 18000 16000 120 100 14000 truth MC extrapolation 80 [0]+[1]*x+[2]*x*x+[3]*x*x*x 12000 truth MC extrapolation 10000 [0]+[1]*x+[2]*x*x+[3]*x*x*x 8000 60 HIJING PbPb s=2.76 TeV HIJING PbPb s=2.76 TeV 6000 40 4000 20 2000 0 0 1 2 3 4 0 0 5 1 2 3 4 ∆η 2000 ×103 2500 2000 truth MC extrapolation 1500 [0]+[1]*x+[2]*x*x+[3]*x*x*x 1000 1000 HIJING PbPb s=2.76 TeV 500 0 0 500 1 2 3 4 5 0 0 F120 1500 ∆η F120 F111 ×103 2500 5 ×10 3 Iterations Extrapolation Truth MC 2000 1500 1000 500 0 |η|<0.1 -2.5<η<-2.3 -1 -0.5 1 ∆η 0 0.5 2 1 1.5 3 2 2.5 3 Iteration 4 5 ∆η Factorial moments of rank 2 and 3 are significantly shifted with respect to the truth value. Piotr Janus, Jakub Kremer (AGH) Application of the HBOM method Kraków, 8.1.2016 9 / 11 Closure test III F101 F100 Factorial moments as a function of distance between bin centers corrected with exp(ax 2 + bx + c) + dd. 20000 140 18000 16000 120 100 14000 truth MC extrapolation 80 exp([0]+[1]*x+[2]*x*x)+[3] 12000 truth MC extrapolation 10000 exp([0]+[1]*x+[2]*x*x)+[3] 8000 60 HIJING PbPb s=2.76 TeV HIJING PbPb s=2.76 TeV 6000 40 4000 20 2000 0 0 1 2 3 4 0 0 5 1 2 3 4 ∆η 2000 1500 ∆η F120 F111 ×103 2500 ×103 2500 2000 truth MC extrapolation 1500 exp([0]+[1]*x+[2]*x*x)+[3] 1000 truth MC extrapolation exp([0]+[1]*x+[2]*x*x)+[3] 1000 HIJING PbPb s=2.76 TeV 500 0 0 5 HIJING PbPb s=2.76 TeV 500 1 2 3 4 5 0 0 1 ∆η 2 3 4 5 ∆η Exponential function reproduces truth value very well in the whole pseudorapidity range. Piotr Janus, Jakub Kremer (AGH) Application of the HBOM method Kraków, 8.1.2016 10 / 11 Summary The general problem of correcting for detector effects in HEP measurements was discussed. The HBOM method was studied on high track multiplicity events generated with HIJING. Factorial moments were choosen as an example observable. Presented method gives promising results. More studies of the fitted function are needed. Piotr Janus, Jakub Kremer (AGH) Application of the HBOM method Kraków, 8.1.2016 11 / 11