Minutes of the Computing meeting held on 15 May 2000

Present: J.Closier, M.Cattaneo, T.Cass, G.Gracia, A.Jacholkowska, M.Needham, J.Harvey, I.Korolko, K.Harrison,
              P.Bartalini, F.Ranjard, E.van Herwijnen, T.Nakada, O.Callot, P.Mato, J.Ocariz, G.Corti, F.Harris, I.Belyaev

Minutes : Florence Ranjard

   1. Geant4 in Gaudi framework (I.Belyaev) (pdf)

• current Geant4 has been  installed by Vanya in $LHCBHOME/geant4/geant4.1.1 in verbose mode, global shared libraries.
• DAWN & DAWNFILE version 3.81 is installed in $LHCBHOME/geant4/DAWN
• DAVID version 1.34a is installed in $LHCBHOME/geant4/DAVID
• Vanya has created 2 CMT packages, released on $LHCBSOFT,  to run Geant4 examples and stand-alone applications:
  • GEANT4/v1r1 is a dummy package used to define all environment variables
  • GEANT4Examples/v1 to run the 6 Geant4 novice examples in a stand-alone way.
• Vanya has created 2 CMT packages to run Geant4 in the GAUDI framework:
  • GiGa a Service which gives access to Geant4 in the Gaudi framework
  • GiGaExamples which gives the same 6 Geant4 novice examples as examples of use of Geant4 in Gaudi framework:
    stand-alone Geant4 applications are well fitted into GiGa schema without any change in code.
• Future GiGa evolution implies:
  • giving the possibility to extract the event record from Gaudi event store as well as the detector description.
  • make the conversion of Gaudi detector description into Geant4 detector description.
  • make the conversion of Geant4 hits into Gaudi hits.
  • populate the Gaudi Event store with Geant4 trajectories.
• Consequence for the user:
  • no user's algorithm deals directly with GiGa service and Geant4 classes.
  • all knowledge of Geant4 is absorbed by a set of specific converters.
  • configuration of Geant4 Physics List and/or cut-offs made by Gaudi via jobOptions service or interactive scripting language.

       It is not yet clear when these improvements will be available and who will work on them.

    2. Bookkeeping improvements: generator parameter table (F.Ranjard)

• Following P.Bartalini's questions during last SICB meeting about the generator parameters table he proposed some months ago,
  Florence had a look at the present status of the bookkeeping:
  • ORACLE tables are filled at end of job with variables set outside the Fortran program.
  • ORACLE tables reflect the program settings not the program itself.
  • to reflect the program itself we need an interface Fortran-ORACLE.
  • It is not foreseen to modify Fortran programs to use such interface , that will be done in Gaudi.
  • In the meantime tables are filled with environment and script variables: we have to define as many variables as necessary.
• Paolo and Florence have worked out a generator parameters table and a way to fill it using the current updating procedure.
  • the generator parameters table should be recognized by a unique key called PID. The table defines the necessary settings for a given event type.
  • The table should be filled by the responsible person (Paolo) before starting a production.
  • The table content should be used to fill data cards necessary for this event type production.
• Some modifications will be introduced in the Runsets table
  • Simulation version replaced with Program Name
  • Reconstruction version removed  (use navigation to find it)
  • beam energy, beam angle, vertex smearing and luminosity moved to another table to be defined.
• User requirements
  • give access to production log files
  • select DSTs made with a given SICBDST version from RAWH made with a given SICBMC version.

    3. A.O.B.

• Release Policy (Flr)
  • Code should be tested thoroughly before giving it to the librarian.
  • If necessary several hundred events should be produced by code writers to make sure that their code is correct.
    • code writers should not expect that the production manager will test their code.
    • Joel and Florence will try to provide scratch disk space for that.
  • Release of a new version should be done once a week preferably on Monday and not on Friday to allow any production to run smoothly during the week end.

• HARP experiment interest in Gaudi (P.Mato)
  • HARP experiment is interested in using the GAUDI framework. They will use GEANT4 for the simulation, so they
    could be interested to participate to the deployment of GEANT4 in GAUDI. Pere will pursue the discussion with the
    HARP collaboration.

• Focus questionnaire (M.Cattaneo)
  • Marco, as LHCb representative, will prepare a response to the FOCUS questionnaire he has sent to experiments,
    as FOCUS secretary.
    The questionnaire is about the RISC decommissioning plans proposed by IT division.
    Marco will circulate the response by e-mail.

===============================================================

next meeting :  29 May  2000 at 14:00 in bldg 1/1-025