HLT-TRD Henner Büsching Fast Track Closeout Meeting, June 07, 2004 Henner Buesching University of Frankfurt ALICE workshop Sibiu 08/22/08
Providing ’physics’ trigger decisions Online event reconstruction and analysis Performance monitoring of the ALICE detectors Online production of calibration data Lossless compression of event data High Level Trigger system
Event Selection relies on: –Processing in analysis steps Serial steps of feature extraction / hypothesis testing Modular analysis chain Events can be rejected at any step if features do not fulfil certain criteria (signatures) Emphasis on early event rejectionEmphasis on early event rejection –Reconstruction in Regions of Interest (RoIs) RoI size/position derived from previous step(s) High Level Trigger Strategy 3
First steps to define triggers Jet Finder (TPC) –Fast seeded cone algorithm based on charged tracks in TPC Single-electron (TRD) –improve statistics of open-charm and open- beauty at ( p T > GeV/c) Di-Muon trigger Open charm 0 / – gamma conversion 4 Merge TPC -TRD tracking
ALICE High-Level Trigger 5
Hardware Status First year setup complete 87 Frontend PCs GB Memory CPU cores HLT Read-Out Receiver Card DDL links 16 Infrastructure PCs All Optical Fibers to DAQ installed
Hardware Status GigaBit Ethernet network operational Interface nodes (2 each) to ALICE online systems - ECS / DCS / Offline
High Level Trigger system DAQHLTDCSTrigger ECS raw event data analyzed events / trigger decisions Mass Storage
Dynamic software data transport framework (Publish- Subscriber) Cluster management (TaskManager) Analysis Software (AliRoot) Software architecture
Components and the Chain 10
11 Monitoring Scheme
12
Monitoring 13
Monitoring at CERN 14
15 Time of Testbeam ONLINE HLT-TRD monitoring was successfully running online in “test modus” at last beam time
16 Time of Testbeam ONLINE
Online Monitoring Tool 17 Powerfull visualization tool for TRD exists (A.Bercuci) Simple HLT online tool Developed by RISE summer student Kurt M. Barry
18
Code consistency A strategic decision: Using offline code in HLT Pro: –Same results offline/online –Offline code well tested –Reliable usability for physics trigger decisions Con: –Potentially slow –Cumbersome optimization 19
20 Getting ready for pp: Challenges Continuity –Guaranteeing operation in constantly ‘improving’ framework environment Stability –Fixing memory leaks Speed –Optimizing and speeding up routines
The Tracking Algorithm clusters
Tracking Algorithm 4 seeding clusters
Tracking Algorithm Riemann fit Linear fit
Tracking Algorithm 2 cut: Finding real clusters around fit
Tracking Algorithm tilted fits + Kalman fits
26 Summary ALICE HLT core system ready for data taking Online event reconstruction and analysis for TRD implemented Performance monitoring of TRD possible Running as it was designed to, but it now relies on the performance of the offline code. Task force working on ’physics’ trigger decisions
Sebastian Robert Bablok, Oystein Djuvsland, Kalliopi Kanaki, Joakim Nystrand, Matthias Richter, Dieter Roehrich, Kyrre Skjerdal, Kjetil Ullaland, Gaute Ovrebekk, Dag Larsen, Johan Alme (University of Bergen, Norway) Torsten Alt, Volker Lindenstruth, Timm M. Steinbeck, Jochen Thaeder, Udo Kebschull, Stefan Boettger, Sebastian Kalcher, Camilo Lara, Ralf Panse (Ruprecht-Karls-University Heidelberg, Germany) Konstantin Antipin, Harald Appelshäuser, Henner Büsching (University of Frankfurt, Germany) Mateusz Ploskon (UC Berkeley, USA) Haavard Helstrup, Kirstin F. Hetland, Oystein Haaland, Ketil Roed, Torstein Thingnaes ( Bergen University College, Norway) Kenneth Aamodt, Per Thomas Hille, Gunnar Lovhoiden, Bernhard Skaali, Trine Tveter ( University of Oslo, Norway) Indranil Das, Sukalyan Chattopadhyay (Saha Institute of Nuclear Physics, Kolkata, India) Bruce Becker, Corrado Cicalo, Davide Marras, Sabyasachi Siddhanta (Cittadella Universitaria, Cagliari, Italy) Jean Cleymans, Artur Szostak, Roger Fearick, Gareth de Vaux, Zeblon Vilakazi (University of Cape Town, South Africa) Credits