Detector description
David Adams
10apr02 1115


Detector description should be provided as a series of layers
which build on earlier layers and add services. Here we
describe such a high-level design.

The goal is to provide a common description that can be shared
by different consumers including simulation, reconstruction
and visualization.

This description must have at least one realization in each
programming language (C++ and possibly Java, FORTRAN, Python and
others) and at least one persistent representation so the
description can be shared between executables.


1. Identification
-----------------

The first level is identification. Each detector element is
is assigned an identity which is used as the key for discovering
properties described below.

Each identifier has a string representation so it can be
understood by humans.


2. Structure
------------

An identified element and is typically associated with a
collection of other elements. The totality of these
associations constitute the structure of the detector.

The simplest description would be a single tree where each
element has one parent and a collection of children. However,
it appears that a single tree cannot satisfy the needs of
both simulation and reconstruction and perhaps not the needs
of their subsystems (e.g. readout and track-finding).

We should decide whether to use a more complex structure
(elements have multiple parents) or have multiple trees.

The (or each) structure provides a means to navigate
between elements, i.e. discover the parent(s) and children
for a given element.


3. Properties
-------------

There are properties associated with each detector element.
Thes include:
  a) shape and size
  b) position
  c) orientation
  d) digitization data (e.g. to convert position to TDC count)
  e) reconstruction data (e.g. to convert TDC count to drift
     distance)
  f) Material properties (density, Lrad, X0)

All of these properties can be accessed the appropriate
identifier as as a key.