Validation of the ATLAS Hadronic Calibration with LAr End-Cap Beam - PowerPoint PPT Presentation

Validation of the ATLAS Hadronic Calibration with LAr End-Cap Beam Tests data Teresa Barillari, MPI M¨ unchen Calor 2008 Pavia, 26 - 30 May 2008 � Introduction � Hadronic Calibration Performances � Linearity and Resolution � Cluster Moment distributions � Outlook on behalf of the ATLAS LAr End-Cap Group: Canada, China, France, Germany, Russia, Slovakia, Spain, U.S.A.

Introduction � In this talk I will present the initial studies done to validate the MC simulation using the full ATLAS hadronic calibration procedure (see talk from G. Pospelov) with the EMEC-HEC-FCal beam tests data � The data shown are π s obtained in the H6 2004 combined beam tests (see talk from P . Strizenec). They correspond to an ATLAS region 2 . 5 < | η | < 4 . 0 , and cover a energy range 40 ≤ E ≤ 200GeV � The MC simulation used in this analysis consists π s obtained using Geant 4 8 . 3 QGSP EMV � To preliminary compare the MC with the beam tests data, the full ATLAS Geant 4 MC has been restricted so to approximately describe the beam tests area � Results are presented for data taken at point D (EMEC-HEC) and and at point H (FCal), see red circles in the right figure Teresa Barillari, MPI M¨ unchen Calor 2008 Pavia, 26 - 30 May 2008 2

Hadronic Calib. Performance: π s at Point D � Plots show the total reconstructed energy over the nominal beam energy for 60 GeV π and 200 GeV π at Point D for EM-scale (red); weighted (green); weighted + OOC (blue); weighted + OOC + DM (black) � Data point D: 60 GeV π 160 h_em Entries 3632 140 • Mean improves in every step: Mean 0.6988 RMS 0.104 Mean of the fit 0.6842 Sigma of the fit 0.1137 after em = 0 . 684 , after weight = 0 . 780 , 120 h_w Entries 3632 Mean 0.7883 after ooc = 0 . 844 , after dm = 0 . 943 RMS 0.1279 100 Mean of the fit 0.7804 Sigma of the fit 0.13 • Resolution remains about h_ooc 80 Entries 3632 Mean 0.8456 constant: after em = 16 . 27 % , RMS 0.1363 60 Mean of the fit 0.8436 Sigma of the fit 0.1358 after weight = 16 . 41 % , h Entries 3632 40 Mean 0.944 RMS 0.1532 after ooc = 15 . 7 % , after dm = 15 . 63 % Mean of the fit 0.943 Sigma of the fit 0.1496 20 • Final deviation from 0 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 E reco / E π = 1 is 5 . 7 % E /E reco π � Data point D: 200 GeV π 900 h_em 800 Entries 17223 Mean 0.756 • Mean improves in every step RMS 0.08915 Mean of the fit 0.7556 700 Sigma of the fit 0.08684 after em = 0 . 756 , after weight = 0 . 869 , h_w Entries 17223 600 Mean 0.8657 after ooc = 0 . 906 , after dm = 0 . 993 RMS 0.09844 Mean of the fit 0.8686 500 Sigma of the fit 0.08773 • Resolution improves in every h_ooc Entries 17223 400 Mean 0.9019 RMS 0.1038 step after em = 12 . 61 % , Mean of the fit 0.9056 Sigma of the fit 0.09061 300 h after weight = 11 . 09 % , Entries 17223 Mean 0.9863 200 RMS 0.1142 Mean of the fit 0.9929 after ooc = 10 . 91 % , after dm = 10 . 39 % Sigma of the fit 0.09878 100 • Final deviation from the 0 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 E reco / E π = 1 is ≈ 0 . 7 % E /E reco π Teresa Barillari, MPI M¨ unchen Calor 2008 Pavia, 26 - 30 May 2008 3

Linearity and Resolution at Point D � Plots show the linearity and the resolution for 40 GeV ≤ π ≤ 200 GeV , data and MC, at Point D. Full (open) red circles are data (MC) at EM-scale. Full (open) black triangles are data (MC) after weighting + OOC + DM corrections 1.1 π /E reco 1 � Linearity at point D: E 0.9 • There is a difference between data ( π ) and MC at 40 and 0.8 60 GeV . At higher energies the 0.7 calibrated data are in fairly good 0.6 agreement with the MC 3 10 × 0.5 40 60 80 100 120 140 160 180 200 E (MeV) π /E 0.24 σ 0.22 0.2 � Resolution at point D: 0.18 • The MC gives an overall better 0.16 0.14 resolution than the data, both at 0.12 EM-scale and after calibration 0.1 corrections 0.08 0.06 3 × 10 40 60 80 100 120 140 160 180 200 E (MeV) π � We know the noise in the data is higher than in the MC. We also know that the lateral shower shape of Geant 4 QGSP doesn’t describe data Teresa Barillari, MPI M¨ unchen Calor 2008 Pavia, 26 - 30 May 2008 4

Energy Density Moment at Points D and H Ecell � Vcell ∗ E cell � Plots show the energy density cluster moment: log 10 ( cells ) for � cells E cell 200 GeV π at Point D (left plot) and H (right plot) 0.045 0.06 Point H, 200 GeV π 0.04 Point D, 200 GeV π Data 0.05 0.035 Data MC 0.03 MC 0.04 0.025 0.03 0.02 0.02 0.015 0.01 0.01 0.005 0 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 3 log (< ρ > (MeV/mm )) 3 log (< > (MeV/mm )) ρ 10 10 � At both points D and H the MC has slightly higher energy density than the data, we can say it reasonable describes the data � The energy density moment is used in the calibration procedure for separation of hadronic from electromagnetic showers Teresa Barillari, MPI M¨ unchen Calor 2008 Pavia, 26 - 30 May 2008 5

Second Radial Moment at Points D and H � Plots below show the sqrt of the second radial moment (lateral spread) for 200 GeV π at points D (left plot) and H (right plot) 0.12 0.045 0.04 0.1 0.035 Point D, 200 GeV Point H, 200 GeV π π 0.08 0.03 Data Data MC MC 0.025 0.06 0.02 0.04 0.015 0.01 0.02 0.005 0 0 0 20 40 60 80 100 120 140 160 180 200 0 20 40 60 80 100 120 140 160 180 200 2 2 <r > (mm)) <r > (mm)) � Even though there is a shift to the right, at point H the MC describes the data. At point D the MC fails to describe the data � The shower model used in MC physics list doesn’t describe the data well Teresa Barillari, MPI M¨ unchen Calor 2008 Pavia, 26 - 30 May 2008 6

Outlook � We have started to validate the ATLAS Hadronic Calibration procedure using the LAr End-Cap Beam Tests data � In the studies shown in this presentation we see that the MC Geant 4 8 . 3 QGSP EMV has problems to describe the data � We need and we are working to have a proper beam tests MC simulation � Once a test-beam MC simulation is available, further improvements for the hadronic calibration procedure (EM-scale, weighting, OOC and dead material correction) can be done � These are the first validation results with LAr end-cap beam tests data obtained so far Teresa Barillari, MPI M¨ unchen Calor 2008 Pavia, 26 - 30 May 2008 7

BACKUP SLIDES

Hadronic Calib. Performance: π s at Point H � Plots below show E reco / E π for π in Point H for EM-scale (red); weighted (green); weighted + OOC (blue); weighted + OOC + DM (black). 300 h_em Entries 9570 Mean 0.7124 RMS 0.1138 Mean of the fit 0.694 250 Sigma of the fit 0.1256 � Data point H: 60 GeV π h_w Entries 9570 Mean 0.7345 200 RMS 0.1167 Mean of the fit 0.7223 • Mean improves in every step, the Sigma of the fit 0.125 h_ooc Entries 9570 150 resolution becomes worse Mean 0.8596 RMS 0.1401 Mean of the fit 0.8564 • Final deviation from Sigma of the fit 0.1396 100 h Entries 9570 E reco / E π = 1 is 6 . 03 % Mean 0.9397 RMS 0.1688 Mean of the fit 0.9347 50 Sigma of the fit 0.166 0 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 E /E reco π h_em 700 Entries 17163 Mean 0.7787 RMS 0.09803 Mean of the fit 0.7798 Sigma of the fit 0.0956 600 � Data point H: 200 GeV π h_w Entries 17163 Mean 0.8126 500 RMS 0.1028 Mean of the fit 0.8173 • Mean and resolution improve in Sigma of the fit 0.09749 400 h_ooc Entries 17163 every step of the corrections Mean 0.8787 RMS 0.1133 300 Mean of the fit 0.8865 Sigma of the fit 0.1031 • Final deviation from the h Entries 17163 200 E reco / E π = 1 is ≈ 7 . 21 % Mean 0.9202 RMS 0.1213 Mean of the fit 0.9279 Sigma of the fit 0.1082 100 0 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 E /E reco π Teresa Barillari, MPI M¨ unchen Calor 2008 Pavia, 26 - 30 May 2008 9

Linearity and Resolution at Point H � Plots show the linearity and the resolution for data and MC at Point H. Full (open) red circles are data (MC) at EM-scale. Full (open) black triangles are data (MC) after weighting + OOC + DM corrections 1.1 π /E reco 1 E 0.9 � Linearity at point H: 0.8 • The calibrated data (for E > 40 GeV ) show a fairly good 0.7 agreement with the MC 0.6 3 10 × 0.5 40 60 80 100 120 140 160 180 200 E (MeV) π 0.3 /E σ � Resolution at point H: 0.25 • In the low energy region the data have a better resolution than the 0.2 MC. In the higher energy bins the 0.15 MC performs better. The dead material correction are 0.1 problematic in this region for both 3 10 data and MC. × 0.05 40 60 80 100 120 140 160 180 200 E (MeV) π Teresa Barillari, MPI M¨ unchen Calor 2008 Pavia, 26 - 30 May 2008 10