1 / 30

T.Yoshioka (ICEPP) , M-C.Chang(Tohoku), K.Fujii (KEK),

Realistic Particle Flow Algorithm for GLD. T.Yoshioka (ICEPP) , M-C.Chang(Tohoku), K.Fujii (KEK), T.Fujikawa (Tohoku), A.Miyamoto (KEK), S.Yamashita (ICEPP), on behalf of ACFA-SIM-J Group. Contents : 1. Introduction 2. Procedure of PFA - Small Clustering - Gamma Finding

talia
Download Presentation

T.Yoshioka (ICEPP) , M-C.Chang(Tohoku), K.Fujii (KEK),

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Realistic Particle Flow Algorithm for GLD T.Yoshioka (ICEPP), M-C.Chang(Tohoku), K.Fujii (KEK), T.Fujikawa (Tohoku), A.Miyamoto (KEK), S.Yamashita (ICEPP), on behalf of ACFA-SIM-J Group Contents : 1. Introduction 2. Procedure of PFA - Small Clustering - Gamma Finding - Track Matching 3. Results 4. Summary The 8th ACFA Workshop@EXCO, Daegu, Korea

  2. Introduction • Most of the important physics processes studied in the future linear • collider experiments have multi-jets in the final state. • → Precise Jet reconstruction is essential. • Since the momentum resolution of the charged particle is • much better than the energy resolution of the calorimeter, we use • Trackers for charged particles • Calorimeters for neutral particles • for Jet reconstruction. • → Particle Flow Algorithm (PFA) • In this talk, we present a performance of the PFA using a full • simulator of GLD named Jupiter. The 8th ACFA Workshop@EXCO, Daegu, Korea

  3. Calorimeter Geometry in Jupiter - Side view Barrel Tower Front : 210cm Endcap Inner R : 40cm Endcap Tower Front Z : 270cm - End view 210 cm 210 cm 40 cm 270 cm - Consists of tower structure. Thanks to Y.Yamaguchi (Tsukuba) The 8th ACFA Workshop@EXCO, Daegu, Korea

  4. Full One Tower ECAL + HCAL Calorimeter Geometry in Jupiter - Side view Barrel Tower Front : 210cm Endcap Inner R : 40cm Endcap Tower Front Z : 270cm - Tower - # of Layers ECAL : 38 HCAL : 130 210 cm HCAL 40 cm 270 cm ECAL Thanks to Y.Yamaguchi (Tsukuba) The 8th ACFA Workshop@EXCO, Daegu, Korea

  5. Full One Tower ECAL + HCAL 6.1λ 27 X0 Calorimeter Geometry in Jupiter - Side view Barrel Tower Front : 210cm Endcap Inner R : 40cm Endcap Tower Front Z : 270cm - Tower - # of Layers ECAL : 38 HCAL : 130 • Cell • Cell Size • EM : 4cm x 4cm • HD : 12cm x 12cm • Can be changed easily. 210 cm HCAL 40 cm 270 cm ECAL Thanks to Y.Yamaguchi (Tsukuba) The 8th ACFA Workshop@EXCO, Daegu, Korea

  6. Energy Deposit in a Cell - Z → qq-bar(q = u,d,s)@91.187GeV. ECAL HCAL The 8th ACFA Workshop@EXCO, Daegu, Korea

  7. PFA Procedure - Current Scheme of the PFA • 1.Small Clustering (collect fired cells with its neighbors) • → Define thrust and broadening for each small cluster. 2. Gamma Finding (Separate gamma/e to hadron) → Small Cluster based. 3. Cluster-Track Matching (Separate charged to neutral) → Calorimeter Cell based. Flow to make PFObject Note: → Cheating method is used for charged hadrons in the Endcap region. → Assuming the remaining clusters be neutral hadrons. The 8th ACFA Workshop@EXCO, Daegu, Korea

  8. Small Clustering (1) Find the “starting cell” among those “fired cell” in ECAL and HCAL which contains the highest energy deposition. (2) From the “starting cell”, find its “fired” neighbor cells to form a small cluster. Neighbor Cells “Fired” = edep>50keV If the cluster energy exceeds 50MeV(before calibration), the clustering is stopped. Starting Cell (Highest energy cell) The 8th ACFA Workshop@EXCO, Daegu, Korea

  9. Small Cluster Distributions - Z → qq-bar(q = u,d,s)@91.187GeV. Number of hits in a small cluster Small Cluster Energy The 8th ACFA Workshop@EXCO, Daegu, Korea

  10. Variables for Small Cluster - Energy-weighted Thrust & Broadening Thrust axis (n) Direction from the center to each hit (Xi) Center of the small cluster Small Cluster - Thrust axis n is defined as to maximize the T. The 8th ACFA Workshop@EXCO, Daegu, Korea

  11. Thrust and Broadening - Z → qq-bar(q = u,d,s)@91.187GeV. Thrust Broadening The 8th ACFA Workshop@EXCO, Daegu, Korea

  12. Gamma Finding Procedure • Fit the longitudinal profile of the energy deposit of • a small cluster as an electromagnetic cascade. • (2) Make a cut on the distance from the nearest track. • (3) Compare energy sum between HCAL/ECAL within a tube. So far, we achieved Efficiency : 60%, Purity : 99%. Details of the gamma finding algorithm/performance are described by next speaker (T.Fujikawa) The 8th ACFA Workshop@EXCO, Daegu, Korea

  13. Track Matching Procedure • Basic Concept : • Extrapolate the charged track and calculate a distance between • a calorimeter hit cell and the extrapolated track. Connect the cell • that in a certain tube radius (clustering). Extrapolated Track • Calculate the distance • for any track/calorimeter • cell combination. HCAL Tube Radius Hit Cells • Tube radius for ECAL • and HCAL can be changed • separately. distance ECAL Calorimeter input position Charged Track The 8th ACFA Workshop@EXCO, Daegu, Korea

  14. Track Matching Procedure - More detail of the procedure… • Sum up calorimeter cell energies within a tube for a track • (Ecluster). • (2) If Ecluster > Etrack x (0.4 x nsigma + 1), stop the clustering for • the track. If not, perform (1) for each track. (We assume the • calorimeter resolution of 40%.) • (3) Widen the tube radius (only a Rstep) and perform (1) and (2) • again. • (4) Perform (1)-(3) until the tube radius reach the max radius (Rmax). Parameters (nsigma, Rstep and Rmax) should be optimized. The 8th ACFA Workshop@EXCO, Daegu, Korea

  15. Distance Distribution - Z → qq-bar(q = u,d,s)@91.187GeV. ECAL HCAL The 8th ACFA Workshop@EXCO, Daegu, Korea

  16. Efficiency/Purity - Efficiency/Purity of Track Matching is checked by cheating method. Efficiency : 86.4% Purity : 82.2% The 8th ACFA Workshop@EXCO, Daegu, Korea

  17. e- e+ Event Display Just in order to get a feeling of current procedure… The 8th ACFA Workshop@EXCO, Daegu, Korea

  18. e- e+ Event Display Z → qq-bar Red : pion Yellow : gamma Blue : neutron The 8th ACFA Workshop@EXCO, Daegu, Korea

  19. Event Display At first, charged hadrons in Endcap region are removed by cheating method. The 8th ACFA Workshop@EXCO, Daegu, Korea

  20. Event Display After Endcap cheating… The 8th ACFA Workshop@EXCO, Daegu, Korea

  21. Event Display In the next, e/gamma like clusters are removed by “Gamma Finding”. The 8th ACFA Workshop@EXCO, Daegu, Korea

  22. Event Display After e/gamma finding… Note Efficiency : 60% Purity : 99% The 8th ACFA Workshop@EXCO, Daegu, Korea

  23. Event Display Then, Then, calorimeter hits near the extrapolated track are collected by “Track Matching”. The 8th ACFA Workshop@EXCO, Daegu, Korea

  24. Event Display Remaining clusters are assumed to be neutral hadrons. The 8th ACFA Workshop@EXCO, Daegu, Korea

  25. Z Energy Resolution Sum up Calorimeter Energy Particle Flow Algorithm The 8th ACFA Workshop@EXCO, Daegu, Korea

  26. Change Calorimeter Cell Size EM : 1cm x 1cm HD : 1cm x 1cm EM : 4cm x 4cm HD : 12cm x 12cm • In the current method, the resolution is not improved so much • by fine segmentation. → Another method should be considered • in this case. (tracking and vertex finding in calorimeter, etc…) The 8th ACFA Workshop@EXCO, Daegu, Korea

  27. Can We Improve More? pi Particle ID of hit cell • Current e/gamma finding • efficiency is ~60%. • → Improve efficiency while • keeping high purity. • There are still ~10% charged • hadrons after track matching. • → Optimize parameters • (Tube radius etc…) : EC cheating e/gamma : e/gamma finding : track matching : remaining hits K p KL n The 8th ACFA Workshop@EXCO, Daegu, Korea

  28. e/gamma Contamination Track matching purity Track matching purity w/o e/gamma • If e/gamma contamination are removed (by cheating), • the track matching purity improve from 82.2% to 92.2%. • → Z energy resolution would also be improved. The 8th ACFA Workshop@EXCO, Daegu, Korea

  29. Track Matching Performance - Collected calorimeter energy minus track energy • Lower tail indicates that • current tube radius is narrow • (not efficiently collected energy). • 1. Optimize parameters. • 2. Energy dependent tube radius? The 8th ACFA Workshop@EXCO, Daegu, Korea

  30. Summary and Future • Realistic Particle Flow Algorithm (PFA) for GLD is developed • and the performance is checked. • - Gamma Finding : Efficiency : 60%, Purity : 99% • (Details will be presented by T.Fujikawa) • - Track Matching : Efficiency : 86.4%, Purity : 82.2% • - Z energy resolution : 40%/sqrt(E) • - Goal : 30%/sqrt(E) of Z energy resolution. • - Treat Endcap properly. • Improve gamma finding efficiency. • Optimize track matching parameters. The 8th ACFA Workshop@EXCO, Daegu, Korea

More Related