130 likes | 135 Views
TPC reconstruction in the HLT. S. Gorbunov 1 and I. Kisel 2,3 ( for the ALICE Collaboration ) 1 Kirchhoff Institute for Physics, University of Heidelberg, Germany 2 Gesellschaft für Schwerionenforschung mbH , Darmstadt, Germany
E N D
TPC reconstruction in the HLT S. Gorbunov1 and I. Kisel2,3 ( for the ALICE Collaboration ) 1Kirchhoff Institute for Physics, University of Heidelberg, Germany 2 Gesellschaft für Schwerionenforschung mbH, Darmstadt, Germany 3Laboratory of Information Technologies, JINR, Dubna, Russia ALICE offline week CERN, October 23, 2008
TPC reconstruction scheme TPC slice 35 TPC slice 1 TPC slice 0 Cluster finder Cluster finder Cluster finder … clusters clusters clusters Slice tracker Slice tracker Slice tracker slice tracks slice tracks slice tracks TPC Global merger TPC tracks Sergey Gorbunov, KIP
Tracking with the Cellular Automaton method: principles birth survival death • “Game of life” - origin of the Cellular Automatons: • A dead cell with exactly three live neighbors becomes a live cell (birth) • A live cell with two or three live neighbors stays alive (survival) • In all other cases, a cell dies or remains dead (overcrowding or loneliness) • Evolution of all the cells proceeds in parallel, generation by generation • The Cellular Automaton method for tracking, principles: • Construction of global thing (track) using only local operations (hit-to-hit linking) • Try to keep combinatorics at the local level • Try to perform calculations in parallel • Gradual complication of the calculations with complication of processed data Sergey Gorbunov, KIP
Cellular Automaton - step 1: Search for neighbours For each TPC cluster find two (up&down) neighbours which compose the best line TPC clusters row 2 • Search for neighbours: • local search for each TPC cluster • cluster-wise parallelism • less combinatorics left for the further processing TPC row row 1 x y z row 0 Sergey Gorbunov, KIP
Cellular Automaton - step 2: Evolution Keep only one-to-one links row 3 • Evolution step: • no combinatorics • cluster-wise parallelism row 2 row 1 x y z row 0 Sergey Gorbunov, KIP
Cellular Automaton - step 3: Creation of track segments Track Endpoint Clusters => Endpoints Endpoint • Creation of the track segments: • each sequence of the neighboring clusters is composed to the track segment. • track segment-wise parallelism • no search, no combinatorics • fitting mathematics • only endpoints are left for the further reconstruction Belongs to row, array sorted in Y Endpoint object Somewhere in memory Track object Endpoint object Sergey Gorbunov, KIP
Cellular Automaton - step 4: Merging of segments Empty link [dead] Merge endpoints One way link [dead] One-to-one link [alive] • Merging segment endpoints - evolution step: • local search for the best neighbor in the closest rows • endpoint-wise parallelism • competition between links ( track length, closeness of the neighbour ) • endpoints are linked one-to-one • fitting mathematics (2 check) Sergey Gorbunov, KIP
Cellular Automaton: example of event reconstruction Step 0: TPC clusters Step 1: Search for neighbours Step 2: Creation of track segments Sergey Gorbunov, KIP
Cellular Automaton: example of event reconstruction Step 4: Merging of segments Result: Reconstructed tracks Sergey Gorbunov, KIP
Cellular Automaton: TPC slice tracker performance Soft track with 7 turns Reconstruction time for the whole TPC (36 slices) : Construction of cells … 0.5 ms Merging of cells ……... 0.3 ms Creation of segments . 0.3 ms Fit of segments ……… 0.1 ms Merging of segments . 1.8 ms Reconstruction performance: All Set (Hits >10, P > .05) Reference Set (All Set + P>1.) Extra Set (All Set + P<1.) Eff Clone Eff Clone Eff Clone Ghost Time [ms] 94.7 29.6 97.9 5.9 94.5 30.6 2.9 3.3 Sergey Gorbunov, KIP
Full HLT TPC tracker performance (Slice tracker + Global merger) Reconstruction performance for the full HLT TPC tracker called from off-line ( AliTPCComparison macro ) [ done by Cvetan Cheshkov ] Sergey Gorbunov, KIP
Cellular Automaton: use of parallel hardware • Hardware possibility for the parallel calculations: • SIMD CPU instructions • multi-threading • multi-core CPU • special hardware (Cell processor) CBM Sergey Gorbunov, KIP
Summary and plans • Summary: • The Cellular Automaton tracker for ALICE HLT has been developed • The tracker shows good performance and speed • It reconstructs all kinds of data: Physics, Cosmics, Calibration events; with and w/o magnetic field. • Running on-line in HLT • Current work: • Investigation of parallel hardware • Speed up of the global merger • Tuning for Pb-Pb collisions p-p event Pb-Pb event in work Sergey Gorbunov, KIP