Status of Emc fast simulation:

o PacEmc package:

  -  PacEmcDigi       (modified from Babar's EmcDigi)

      private:
        // Data members
        double _energy;
        const TwoCoordIndex* _tci;

        double _thetaReal;
        double _phiReal;
        HepPoint *_where;


     ** Babar's EmcDigi has other information such as time and calibration.


  - PacEmcCluster      (modified from Babar's EmcCluster)

      private:
        // Members
        std::vector< PacEmcDigi* > *_pacEmcDigis;
        unsigned _nbumps;
        std::vector< GTrack > *_gtracks;

        // Cachable data 
        mutable bool _energyValid;
        mutable double _energy;
        mutable bool _whereValid;
        mutable HepPoint* _where;

        mutable PacEmcClusterEnergySums* _theClusEnergySums;
        mutable PacEmcXClMoments* _theClusXClMoments;


  - PacEmcClusterEnergySums

        for calculating e1, e9, e25, e1/e9, e9/e25.

  - PacEmcXClMoments

        for calculating secondMoment, lat, zernike moments.

  - PacEmcClusterMeas

      class PacEmcClusterMeas : public PacMeasurement

    has one method:  

        void createCluster(const PacSimHit& hit, PacEmcCluster *cluster) const;

    in which it simply request for a PacEmcShowerLib from the global environment
    and sample a cluster given a PacSimHit:


     const PacEmcShowerLib* showerlib= Ifd< PacEmcShowerLib >::get(gblPEnv, SOMEKEY );
     cluster= showerlib->sample_cluster(&hit);


  - PacEmcShowerLib

      The shower library interface class, providing pure virtual methods which
      are the interface to the underlying shower library.



o Design of the shower library storage:

  - Use TTrees stored in root files.

  - Each TTree contains N entries, each represents an EMC cluster.

  - Each TTree represents cluster samples in a certain bin of
    momentum/[other binning quantities such as incident angle] for each
    particle type :  will have a lot of trees.

  - A cluster is represented by theta/phi local coordinates (integers) and
    energies, each represents the crystal location relative to the crystal
    intersect with the GTrack.

    The actual implemetation can either be a new class (e.g., TCluster) that
    has three  std::vector<>  member lists,   or a TClonesArray of a crystal
    class (e.g., TCrystal), which has three numbers as its members, or
    simply three parallel branches, just like what we got when using 
    HTValOrderedVector to create an ntuple with beta.


o At the start up of the program, we have to find all trees and create a map
  so that for a given PacSimHit we can find the corresponding tree, but we 
  don't want to load the tree until we want to create a cluster.