TPC performance with HLT clusters Markus Khler Gesellschaft fr - PowerPoint PPT Presentation

TPC performance with HLT clusters Markus Köhler Gesellschaft für Schwerionenforschung, Darmstadt

Motivation Readout data volume of different detectrors from run 138442 (LHC10h) Bandwidth is a limitating factor for Pb-Pb data taking 2

Data Compression central Pb-Pb event ~ 80 MB Assume 200 Hz central ~ 16 GB/s Bandwidth limit ~ 4 GB/s A data compression factor of ~4-6 is needed in the TPC 3

In this talk What is the effect on the TPC reconstruction performance when using HLT clusters as input? From the information presented in this talk the usage of HLT clusters for TPC reconstruction is acceptable. 4

Compare offline reconstruction with raw and HLT input (clusters) Observables ● Agreement on cluster level ● Agreement on track level ● DCAr vs pT ● DeltaPhi/SigmaPhi (pT resolution) ● dE/dx performance Input data sets ● Pb-Pb data 2011 (Run 166532) ● Monte Carlo for pp and Pb-Pb Reconstruction and QA (locally at GSI) ➔ AliRoot Release branch 5-01 5

Number of Cluster per Track Pb-Pb 2011 Run 166532 ➔ To optimize pT resolution single pad clusters are removed from tracking with HLT clusters Same track cuts imply (e.g. #clusters) imply different amount of tracks 6

Number of Cluster used for dEdx To optimize dE/dx performance single pad clusters are used for dE/dx calculations Pb-Pb 2011 Run 166532 7

#cls/#clsFindable Pb-Pb 2011 Run 166532 ➔ When removing single pad clusters less clusters per track are found (has been shown in previous TPC meetings) 8

Energy loss in TPC Pb-Pb 2011 HLT offline Run 166532 9

dE/dx separation with OROC Pb-Pb 2011 Run 166532 Electron-Pion Electron-Pion Separation = 6.84 +/- 0.06 Separation = 6.87 +/- 0.06 momentum Separation are comparable for HLT and offline 10

Comparison of global tracks to constrained TPC tracks DeltaPhi/SigmaPhi HLT offline Pb-Pb 2011 No bias due to HLT cluster Run 166532 11

DCAz Not enough tracks to calibrate optimal! (Only about 700 Pb-Pb MB events so far) 12

DCAr vs different quantities Problems of last week solved ! Pb-Pb 2011 Run 166532 13

DCAr resolution vs pT Pb-Pb 2011 Run 166532 Corresponds to pT resolution in Monte Carlo 14

pT resolution from MC pp Pb-Pb Pb-Pb HLT (Different pT range for pp and Pb-Pb) High-pT enriched data samples for MC studies in pp and Pb-Pb 15

Summary ● The TPC performance for HLT and offline cluster finders was presented ➔ For 2011 Pb-Pb data (Run 166532) ● Cluster variables ● Track variables ● DCAr vs pT (corresponds to pT resolution) ● dE/dx separation (for OROC only) ● Many more observables were shown ➔ For high-pT samples Monte Carlo in pp and Pb-Pb ● pT resolution ➔ Not shown here : LHC10e, LHC10h, LHC11c (data) ➔ Not shown here : pp MB, Pb-Pb (central) Many more parameters are analyzed 16

Conclusion The usage of HLT clusters is acceptable for TPC reconstruction The High Level Trigger cluster finder can significantly increase the collectible statistics of ALICE in Pb-Pb data taking 2011 17

Involved people Artur Szostak, Jacek Otwinowski, Jochen Thaeder, Marian Ivanov, Markus Koehler, Michael Knichel, Matthias Richter, Sergey~Grobunov, Timo Breitner, Thorsten Kolleger, Torsten Alt, Weilin~Yu, Alberica Toia 18

Backup (DCAr vs pT) offline HLT 19