LCD Weekly
Geometry Discussion
01/25/2005

Attendees:

Ron Cassell (RC)
Norman Graf (NG)
Tony Johnson (TJ)
Willy Langeveld (WL)
Jeremy McCormick (JM)
Max Turrissimo (MT)
Dennis Wright (DW)

[Following is from notes by NG and JM.]

---

-compact description (CD)

  -full detector description from this format

  -ideas f/ old detparms.dtd and JM's example

  -expression evaluation, by default to double

  -type="SomeDetElem"

  -implement DetectorType to handle a detector element

  -detector element identifiers

    -id, name, type

-CD converts to various target formats...

  CD ------ LCDD
  | \
  |  \
  |   \
  |    \
  RECO  GODL

  -other configs not ruled out (NG)

-TJ has been working on a CD of SDJan03 and Java code to parse it
and translate to LCDD.  This file has been successfully parsed
using the SLIC simulator package and a HepRep was produced.

-example code of CD in XML was distributed and critiqued

-CVS package for convertor: "GeomConvertor" in LCD CVS

-Materials listed explicitly.  Common ones will migrated to standalone
file.  Or elements can migrated to code via WL's matprop package.

-Several types of detectors have been identified, essentially taken
from hep.lcd classes.  Content of detector elements is unrestricted, 
except that the parsing driver needs to know the type in order to 
delegate to the correct subscriber.

-The "type" attribute is a candidate for promotion to an element, for which
a schema could be defined. (NG)

-action items 

  -Definitions of segmentation and hit ID assignment
  
  -insertion of arbitrary detector support structures using CD

    -need to insert GDML arbitrarily for non-sensitive supports and 
     (possibly) asymmetric readout structures

  -flat detector hierarchy with placement into world; does it impede Geant4?

-geo info should all come from one file (WL)

  -common format read by sim and reco

-how to deal with levels of detail? (WL)

  -currently, no provision for it

  -could be handled in GODL driver which handles calorimeter types
  
-no concrete definition for the reco file format (different from CD?)

  -use experience gained from usage of .INI files and hep.lcd reconstruction code
   to come up with necessary info

-GDML is complete? (DW)

  -addition of SDs, regions, fields = LCDD or "GDML++" (JM)

-tracking vs. cal region in CD

  -tracking regions hard to do generically and automatically (RC)

-Lelaps allows overlaps (WL)

-use grid test for overlaps (WL)

-booleans expensive in G4 because of implementation details (WL)

-example algorithms solicited (JM)

  -calorimetry

    -clustering

  -tracking

    -track finding

-get working 1st (NG)

-G4 materials

  -composites are mass, volume? (WL)

-concerned with multiple file formats, interfaces (JM)

  -unlike Gaudi's consistent interface across the framework (JM)

  -specific codes for certain tasks (NG)

  -get working first, then change/refine interfaces and data formats (TJ)

-suggest to define goals for LCWS in March and Snowmass in August (JM)

  -ID LCWS goals at least (WL, NG)

-Steve Magill at SLAC, ~Feb 15th to August (Snowmass)