- Torre

Subsystem doc suggestions. Include...

- up to date engineering drawing(s) of the detector to which the
  simu model can be related
- thoughtfully scaled BBSIM plots of the simu model showing how and
  to what detail level the engineering design has been implemented
  - enough info to convey how well the model agrees with current
    design; in what respects is improvement needed
- outline of .db file structure that imports the engineering parameters into
  the simulation
- brief description of the algorithms/approach used to build the model from
  the db info
  - motivation for choices and compromises made
    - are there aspects of the model for which Geant3 lacks adequate
      volumes for their proper description (which will be improvable
      with Geant4)?
    - if MANY has been used, why was it used?
    - say why you wrote your own Cerenkov photon generation/propagation
      rather than using Geant3's
  - limitations, capabilities of the implementation
    - can ultimately needed misalignments of components be introduced
      without major design changes?
- describe hit generation
  - in GUSTEP, when and how do you generate a hit?
  - what information goes into the hit, where does it come from, and
    what's the structure of the hit?
  - typical hit counts per track
- describe digit generation
  - describe generation of digi(s) from hit(s); incorporation of new
    information (eg. calibrations, resolutions, noise)
  - brief description of the digi implementation/context (Framework
    integration, language)
  - structure of the digi, referencing between hits and digis
  - this is of course where much development activity is going on,
    new work and refinement/migration of earlier work; describe the
    evolution, future plans, and requirements driving them.
- cite one or two examples of applications of the simulation (typically,
  detector design studies) that have driven the level of detail found
  in the present implementation
- we are moving from an era in which simu requirements were driven primarily by
  detector design studies, with their associated requirements and acceptable
  shortcuts, to an era in which the simulation will serve as the data
  source for the development of reconstruction and analysis, and for
  increasingly refined physics analyses, with a new set of more stringent
  requirements, eg. in response modelling and material accuracy.
  What new requirements do you face and how are you approaching their
  implementation? How do timescales (completion by / needed by) match?
- are there application domains clamoring now for capabilities/detail/
  response model features NOT in the present implementation?
- modelling of imperfections:
  - geometry: to what extent can misalignments be introduced in the model;
    to what extent do they need to be? [my answer to the latter would be,
    if the reconstruction will be tasked to measure a misalignment with
    data, the simulation should be able to introduce a known misalignment
    to test its correct reconstruction]
  - response: what calibrations and response model details are foreseen
    to be incorporated?
- how do you test and validate a new version for release
- do inadequacies elsewhere in the software handicap your simulation
- few words, one transparency on Aslund geometry generation; probably
  to be used in the Aslund section, not your talk
- use of BBSIM to generate parametrizations for Aslund
- plans (if any at present!) for Geant4 migration
- manpower availability, shortcomings now and in the future