Docstoc

BES III Software

Document Sample
BES III Software Powered By Docstoc
					BESIII Offline Software Overview

        Weidong Li & Yajun Mao

       BESIII Collaboration Meeting, Beijing
                 12 January 2006
          Infrastructure and Environment
 Underlying framework
         GAUDI (originally developed by LHCb)
 Simulation
         GEANT4
 Other external LIBs:
         CERNLIB, CLHEP, ROOT, AIDA, XercesC, GDML …
 Database: PGSQL/MySQL
 Software configuration management
         CMT and CVS
 Computer language: C++ (BESII legacy code written in Fortran )
 Operation system: Redhat 7.3/gcc3.2
 Reused code from Belle, BaBar, ATLAS,GLAST …

12/01/2006                        W.Li and Y. Mao                  2
                                 Organization
              BES III Software                     Institutes participating
                                                        Institute of High Energy
                                                         Physics
                                                        Peking University
Framework Core            Calibration &
   Software                Database                     Shandong University
    Weidong Li                   Jinfa Qiu              University of Science and
                                                         Technology of China
Reconstruction             Simulaton                    Hunan University
                                                        Zhejiang University
     Zepu Mao              Huaimin Liu
                                                        Joint Institute for Nuclear
                                                         Research, Russian
Physics Analysis
   Software
    Kanglin He
 12/01/2006                            W.Li and Y. Mao                                 3
        Generator
                         Event Data
                                 HEPEVT Data
        HEPEVT                           Kinematics information only
         Data                    RAW Data
                                         Delivered by DAQ for reconstruction
     G4 Simulation                       Byte stream format
                                 Simulated Event Data
                                         Contain digits, hits and other MC truth
                 Simulated                information
RAW Data
                 Event Data              Ascii file format
                                 REC & DST Data
                                         Reconstructed Data is event data written
      Reconstruction                      as output of reconstruction procedure
                                         DST Data is a reduced event
                                          representation suitable for analysis
        Rec & DST
           Data                          Both in ROOT format
12/01/2006                    W.Li and Y. Mao                                    4
          Event Data Conversion Services
Real data flow

                           T1
                                                                   Persistency
Apparent data flow
                                                                     Service
                                   Data T1
                     Algorithm                                       AsciiFile
                                   Data T2, T3                                   Ascii Dump
                         A                                           CnvSvc
         T2

                           T3
                                    Data T2
        Algorithm                                Transient
                                    Data T4                           RawData
            B                                                                     RawData
                                                 Data Store           CnvSvc

        T4                       Data T3, T4
                 Algorithm
                     C
                                   Data T5                            RootDst    REC&DST
                                                                      CnvSvc       Data
                     T5


 12/01/2006                                      W.Li and Y. Mao                            5
                        Detector Description
 Based GDML (Geometry Description Markup Language), a kind of XML developed
  by GEANT4 group.
 Peking University group expanded the GEANT4 Schema and developed a new
  ROOT Schema for BESIII applications.
 GDML Detector Description has been used for simulation, event display and
  MUC reconstruction.
                               Geometry,
                                Materials,
                               Alignment…


                             XML description




              Classes            Classes                Classes
         for Simulation      for Reconstruction     for Event Display
 12/01/2006                     W.Li and Y. Mao                         6
             Access to Geometry Data

              MDC       TOF           EMC      MUC      Application
             GeomSvc   GeomSvc     GeomSvc    GeomSvc   Layer


                                                        Data Access
                                                        Layer
              MDC        TOF ObjectsEMC        MUC
                         for Reconstruction
              DAO       DAO           DAO      DAO



                           Geometry,
                           Materials
                          Database
                         XML description


12/01/2006                  W.Li and Y. Mao                       7
             Event Display Tool: BesVis




                     Based on ROOT, OpenGL, X3D and XML
                     Support both 2D and 3D view
                     Operations and controls available through
                      menu and toolbar items
                     First version was released in December 2005.
12/01/2006              W.Li and Y. Mao                           8
              Simulation Integration
 BES has ~30 event generators written in Fortran
 Use C++ Hepevt_Wrapper to access the kinematics information
  generated by the generators

                    Simulation            BOOST (BESIII Objected
                                           Oriented Simulation Tool)
               ①                           was originally developed in a
 Generator                                 independent framework.
                                          Integration with BOOST
                ②                          simulation is based on
                       Transient           ATLAS/Athena software.
Simulation            Event Store
                                          Currently both BES
                                           generators and BOOST have
                                           been integrated with the
                                           offline framework.
12/01/2006                   W.Li and Y. Mao                           9
                         Calibration Status
   Framework is based on GLAST calibration infrastructure

                                calibUtil interface
                                   Write/register                  Signifies
                                      Search                       “uses” or
                                       Read                        “accesses”


             Metadata (persistent)                       Data (persistent)

                                                      Dead chns CAL calibs
                MySQL rdbms
                                                       (XML)     (ROOT)


   Status
        Developed MDC, TOF and EMC calibration prototype algorithms
        Defined calibration data for each sub-detector
        Reconstruction algorithms are able to retrieved calibration constants
         through the framework.
12/01/2006                             W.Li and Y. Mao                           10
        Drift Chamber Software Status
                 Simulation
                    Stereo cells are described by G4TwistTube.
                    Digitization is based on parameters from test
                     beam data.
                 The calibration algorithm contains correction to
                    X-T relation, spatial resolution,T0, wire position,
                     time walk.
                 Reconstruction algorithms are: TrkReco and
                  MdcPatRec. For single 1 GeV muons, both of
                  them obtain:
                    Efficiency: ~99%
                    Spatial resolution: ~110 μm

                 dE/dx resolution: ~6% consistent with detector
                  design report.

12/01/2006           W.Li and Y. Mao                                11
             TOF Software Status
                          Simulation
                               Two-layered structure for Barrel
                               Digitization takes into account:
                                light production, light propagation
                                and PMT response etc.
                               Intrinsic time resolution: 85 ps

                          Calibration & reconstruction
                               Calibration algorithm includes
                                time correction, effective velocity
                                and attenuation calibration.
                               Reconstruction results have been
                                used in particle identification


12/01/2006         W.Li and Y. Mao                                 12
       EM Calorimeter Software Status
                                                    Simulation
                                                        End-cap is described by G4IrregBox.
                                                        Full simulation of readout electronics
                                                         is included in digitization.

                                                    Energy calibration
                                                        Digit calibration: converting ADC
                                                         channels into energy.
                                                        Cluster calibration: correcting non-
                                                         linearity of readout signals.

              E /E=[(p0/E)  (p1/ 4 E)  p2]%      Reconstruction
                                                        ΔE/E = ~2% @1GeV
                                                        σz =   0.6 cm @1GeV


12/01/2006                                 W.Li and Y. Mao                                   13
        Muon Counter Software Status
                                      Simulation
                                           RPC as sensitive
                                            detector
                                           Detailed to each read-out
                                            strip
                                      Reconstruction
                                           Tracking algorithm is
                                            seeded by the tracks
      Window     Fired                      extrapolated from MDC.
                 strips
                                           Searching for hits gap by
                                            gap within predefined
                                            windows.
                                      Reconstruction
     Ext track                         efficiency: ~ 99% @
                                       1GeV
12/01/2006                W.Li and Y. Mao                            14
     Physics Analysis Software Status
   Prototype software for physics analysis
         Analysis event data model
         Analysis event builder
         Kinematics fitting
         Secondary vertex finding
         Particle identification


   Physicists already start checking physics performance.




12/01/2006                          W.Li and Y. Mao          15
      Online Event Selection Software
                                                    Reconstruction
                 StepSequencer
                                                      algorithms

                                   Algorithm        Reconstruction
                                  sequences             result


                                                      Consistent      N
                                                                          Reject
                                                      with criteria
 Share the same framework with offline.
                                                         Y
 Fast reconstruction algorithms shall be       N       End of
  executed step by step according to the                AlgSeq
  predefined sequences. In each step,
                                                         Y
  event is rejected if it does not satisfy
                                                         Accept
  any selection criteria.
 Fast reconstruction algorithms and steering prototypes that control
  event selection have been developed and tested in the online
  environment.
12/01/2006                    W.Li and Y. Mao                                 16
                          Release Status
   BOSS alpha release
        integration with BOOST simulation ( done )
        Running the reconstruction chain for MDC, TOF, EMC and MUC can be
         achieved in the BOSS Gaudi framework. ( done )
        EDM for both Raw Data and reconstructed data developed. Completion
         of fully functional Event Data I/O. ( done )
     Status: published on 30th of May, 2005



   BOSS beta release
        Simulation and reconstruction chain functions reasonably well and can
         successfully process thousands of events. ( done )
        Common physics analysis tools and utilities available. ( done )
        Physics study in BOSS can start.
     Status: published on 17th of November, 2005

12/01/2006                         W.Li and Y. Mao                               17
 Progress since 2005 Annual Meeting
 Analysis software development
        Analysis event builder (done)
        Particle identification (done)
        Secondary vertex finding (done)
 Calibration infrastructure and algorithm prototypes (done)
 Event Display Tool: BesVis (done)
 Migration to GDML detector description (done for MUC )
 Performance studies (ongoing)
        Physics: efficiency and resolution
        System: algorithms’ benchmarking and profiling, memory leakage etc.
 Cavern backgrounds and realistic detector simulation (ongoing)


12/01/2006                           W.Li and Y. Mao                           18
             Areas where work is needed
 Detector alignment, dead/hot channel services
 Magnetic field service
 Parameters database
 Simulation of Level1 hardware trigger
 Mixing of background events
 Migration to Gaudi v16r4/LCGCMT_35 that has been tested in our
  target operation system (SLC3).
 Analysis software will still evolve based on experience obtained from
  the current prototype.
 Physics study for online Event Filter
 Data Challenges


12/01/2006                    W.Li and Y. Mao                         19
                         Data Challenge (1)
   Motivation
        To understand the BESIII computing environment
                CPU, storage and network requirements, system performances, bottle
                 necks, etc…
        To check and validate the BESIII software

   Iterate on a set of Data Challenges of increasing complexity
        Running ‘simulation’, ‘reconstruction’ and ‘analysis’ on a large scale
        Testing calibration procedures
        Adding trigger simulation and event filter algorithms
        Producing physics results




12/01/2006                            W.Li and Y. Mao                           20
                       Data Challenge (2)
   It should include
        Subsystem tests: simulation, calibration, reconstruction and
         analysis
        Full offline chain test: Generationsimulationdigitization
         background mixing reconstructionanalysis
        Trigger/Event filter/Offline chain test
        Stress tests for scaled systems (10%, 25%…)




12/01/2006                           W.Li and Y. Mao                    21
                             Summary
   A tremendous amount of progress in 2005
        Simulation is in good shape and people are focusing on more
         realistic digitization procedures.
        Calibration Infrastructure and prototype algorithms have been
         developed.
        Reconstruction algorithms are being tuned for better performance.
        Physics analysis prototype software are also developed and are
         being used in physics studies.

   Data challenge and software performance optimization are
    two major tasks this year.




12/01/2006                      W.Li and Y. Mao                           22
12/01/2006   W.Li and Y. Mao   23

				
DOCUMENT INFO