1. Aspects of LAr Reconstruction S. Rajagopalan ATLAS Week June 4, 2001

2. ATLAS Week, 4 June 2001LAr Reconstruction Overview  Calorimeter Reconstruction in ATRECON  J. Schwindling et. al.  Successfully used in Physics TDR  Now used as a benchmark for testing new code  Migration to OO  Work began in October 1999  Primary people involved today are: (~ 5 FTE for 2001) F. Lanni, H. Ma, S. Rajagopalan, K. Yip (BNL) A. Radu, M. Wielers (Triumf) J. Collot, F. Ohlsson-Malek, S. Albrand (Grenoble) P. Loch (Arizona) J. Schwindling, J.P. Meyer (Saclay) E. Frank, A. Gupta, F. Merritt (Chicago) : Tile

3. ATLAS Week, 4 June 2001LAr Reconstruction Major Milestones Met  Early design in PASO: Jan. 2000  Migrate to Athena: May 2000  LAr Reconstruction used as a test bed for early Athena  First application software to successfully migrate to Athena  3 working days at LBL  First Common Calorimeter InterfacesOct. 2000  QA review of then available components Dec. 2000  S. Albrand (Grenoble)  Combined Reconstruction (egamma) Jan. 2001  Process GEANT4 LAr Hits (Root Objects) Mar. 2001  Lund ReleaseJune 2001  Establishing most of the reconstruction chain  From G3/G4 Hits to Particle Identification

4. ATLAS Week, 4 June 2001LAr Reconstruction Algorithms & DataObjects  The design/packaging of the LAr reconstruction programs follow the basic principle of the Athena Framework Data Object eg: A collection of cells Data Object eg: A collection of clusters Algorithm(s)Clusteringalgorithms + associated corrections Data Objects reside in the Transient Data Store (TDS). Algorithms produce Data Objects and record them in the TDS. If reading/writing a DataObject from/to disk (persistent media), a converter needs to be provided.

5. ATLAS Week, 4 June 2001LAr Reconstruction Transient Event Store Digits Data Flow D D D c c c C C C Cell Reconstruction Cells Cluster Reconstruction Clusters Analysis (ntuples)

6. ATLAS Week, 4 June 2001LAr Reconstruction LAr Data Classes  Data Objects proposed/implemented in March 2001  J. Collot et. al.

7. ATLAS Week, 4 June 2001LAr Reconstruction LAr Algorithm & Data Objects Algorithms/(Packages)Data Objects G3/G4LArHit DigitizationLArDigit ROD packageLArRawChannel Cell ReconstructionCaloCell Tower ReconstructionCaloTower Cluster ReconstructionCaloCluster egammaRecegamma tauRectauObject JetRecjetMissingET

8. ATLAS Week, 4 June 2001LAr Reconstruction Data Access by LAr Algorithms  LArHit: Describes the G3/G4 LAr Hits  Collection of LArHit read from persistency (Hong Ma)  using “Converters” in Athena  Collection made available in the Transient Data Store (TDS) for algorithms to process. GEANT3 Zebra File ZebraCnv RD Event LArHitZebraCnv LArHitCollection GEANT4 Root File LArHitRootCnv Work with Objectivity in progress, but not yet used by LAr

9. ATLAS Week, 4 June 2001LAr Reconstruction Digitization  Digitization Package: (F. Lanni, J.P. Meyer, J. Schwindling)  Emulation of the FEB in the Athena Framework:  Uses a simple gain selection mechanism  Input from G3 Hits  Additional Inputs: Gain, waveform profile, noise level  Hardwired in code, needs to migrate to a Database.  Output: LArDigit containing 5 samples per channel  Work in progress, first prototype is available.  Adding Pile-up in Athena (M. Wielers)  Cannot read multiple events in Athena at present  Can read old Zebra pile-up tapes (E Matrices) in Athena  Signal+Pile-up events produced for “physics TDR” (Single electrons & photons, H , jets)

10. ATLAS Week, 4 June 2001LAr Reconstruction Cluster Reconstruction on Pile-up Events

11. ATLAS Week, 4 June 2001LAr Reconstruction Pile-up comparison with ATRECON

12. ATLAS Week, 4 June 2001LAr Reconstruction  Kin Yip (+ B. Cleland, F. Lanni)  Compute E, t,  2 from LArDigit (5 samples) ROD Package in Athena Geant3 TestBeam Data DigitizationLArHit ROD Package LArDigit LArRawChannel

13. ATLAS Week, 4 June 2001LAr Reconstruction Analysis of 50 GeV photons at  =0.3 through Digitization & ROD packages

14. ATLAS Week, 4 June 2001LAr Reconstruction Cell Reconstruction  Several collections of CaloCell in Transient Data Store  Collections can be keyed by regions  Collections can be keyed by algorithms producing them  Downstream algorithms (Clustering/Particle ID) can simply request the CaloCell collections by specifying the key Geant3 G3 Cell Builder CaloCellLArHit Digitization LArHit ROD CellBuilder CaloCell LArDigitLArRawChannel CaloCell TileReconstruction

15. ATLAS Week, 4 June 2001LAr Reconstruction Tower Reconstruction C C C CaloCell Collection TowerBuilder T T T CaloTower Collection T T T TowerBuilder Configuration A  R = 0.025 x 0.025 Configuration B  R = 0.1 x 0.1 Key B Key A

16. ATLAS Week, 4 June 2001LAr Reconstruction Cluster Reconstruction  Clustering Algorithms  Sliding Window (for EM and Hadronic Clusters)  Cone Algorithm (Fixed cone size around hottest cell)  Nearest Neighbour Algorithm  Clustering and needed Correction algorithms can be dynamically loaded CaloClusterMaker Sliding Window EnergyCorrection PositionCorrections T T T CaloTower Collection CaloCluster E, ,  Sampling……list<CaloCell*> C C C CaloCell Collection

17. ATLAS Week, 4 June 2001LAr Reconstruction Comparison to ATRECON

18. ATLAS Week, 4 June 2001LAr Reconstruction Combined Reconstruction egamma Rec egamma Rec ( JBH, H. Ma, S. R., A. R., M. W.) First version provides Cluster-Track match, Shower Shape, Isolation… (missing conversions) Output in standard ntuples CaloCluster,SimpleTrack egamma Jet Rec Jet Rec ( A. Gupta, F. Merritt, P. Loch, M. Wielers) KT Algorithm implementation. Most work by Tile, LAr beginning to join in the effort CaloObject Jet tau Rec tau Rec (L. Vacavant, I.H., F.P., S.R.) First version in 2.0.0 (Lund) based on Phys. TDR cuts EM Radius, EM Strips, Isolation, # tracks Output in standard ntuples CaloCluster,SimpleTrack tau Object Missing ET Missing ET (B. Caron) First version for Lund Release based on CaloCell Energies only CaloCell MissingET

19. ATLAS Week, 4 June 2001LAr Reconstruction User Feedback/Validation  Quality Assurance : (S. Albrand)  Some of the recommendation already implemented  Extensive comparison between Atrecon & Athena  Fairouz studies of B  J/   (   ) http://isnwww.in2p3.fr/atlas/fairouz/physics/athena.html  Statistically same, per event differences being understood NEW  Validation of software requested on pre-selected physics processes by NEW users to Athena.  Z  eeM. Losty (TRIUMF)  H   E. Tournefier (Grenoble)  tau analysisF. Paige, D. Cavelli  Jets, MissingETB. Caron, J. Pinfold

20. ATLAS Week, 4 June 2001LAr Reconstruction Recent Plots

21. ATLAS Week, 4 June 2001LAr Reconstruction  ->  analysis  ->  analysis (F. Ohlsson-Malek)

22. ATLAS Week, 4 June 2001LAr Reconstruction Shower Shape Comparison (A. Radu, M. Wielers)  Atrecon-Athena in EM Sampling 2 (A. Radu, M. Wielers)  R = E(3x5)/E(7x7)  w = width (3x5)  E = 20 GeV e -

23. ATLAS Week, 4 June 2001LAr Reconstruction Other Details  Particle ID strategies  IKinematic Interface  Proposal by ATLFAST group needs review/adoption  Event Filter requirements  Adoption of new framework features  StoreGate, Data Links, Persistency  Using Detector Description in Reconstruction  Hardwired in code at present  work in progress to read from DB (MySQL, Objectivity)  Persistency support (Objectivity, Root)  Documentation: http://atlas.web.cern.ch/Atlas/GROUPS/LIQARGON/software/Reconstruction/

24. ATLAS Week, 4 June 2001LAr Reconstruction Conclusion  A central framework that is evolving to provide robust support  The reconstruction design has been built over this framework  Much of the ‘Fortran’ code has been migrated.  Validation ongoing, but results are promising.  It now paves the way for work in:  optimizing and developing new algorithms  Physics and Detector performance studies