Simulation Review
- Outline
- Simulation Code Road Map
- Simulation Code Inventory
- Plans
- Cast of Characters
- P.B.Armstrong
- J.R.Bogart
- G.R.Bower
- R.Cassell
- R.Dubois
- D.L.Flath
- A.S.Johnson
- A.P.Waite
PPT Slide
Simulation Code Inventory (SLAC)
l Fully implemented (including PIC)
Gismo
- Still working with a copy taken over a year ago. The GLAST and LCD versions have diverged (architecturally at least) since then!
- Our improvements/modifications
- Remove generation truncation limit
- Fix assorted PDG errors
- u,d quarks swapped
- Higgs renumbered from 9999 to 25
- Correct some B baryon masses
- Remove miscalculated looping limit (essential for background overlay)
- Outstanding
- Fix NaN problem in optimized Linux code (EGCS 1.1 compiler?)
- Fix debug problem in optimized DEC/OSF1 code (EGCS 1.0.3 compiler?)
GismoApps
- Fixes/improvements incorporated
- Add a VXD cryostat
- Fix calorimeter layer encoding to avoid overflows
- Add north/south encoding for tracking hits
- Add a version number to the ASCII output file
- Fixes/improvements still required
- Realistic magnetic field maps
- All GismoApps code now stored/controlled by CVS
- GismoApps being used as a test-bed for
- Development Environment Controller (DEC) testing
- XML testing
- Serial Input/Output (SIO) testing
Root and RootApps
- Joanne Bogart working on introducing XML
- ‘Keepers of the Flame’ are Masako Iwasaki and Homer Neal
- Homer has Root/RootApps running on a DEC/OSF1 platform
- Please see Masako’s talk for further details
Geant and GeantApps
- Version 4.0.1 of Geant was released 19 July 1999
- Dependence on commercial software
- No longer dependent on RogueWave (good)
- Only persistence model fully supported is Objectivity (unfortunate but not fatal)
- Dependence on free software
- CLHEP
- A variety of visualization schemes
- Cygnus Solutions ‘Cygwin’ environment (Windows/NT only)
- Trying to persuade SCS to support Geant centrally
- No LCD applications written yet!
CLHEP (for Gismo) and CLHEP
- Class Library for High Energy Physics (CERN product)
- Three versions exist at SLAC
- CLHEP for Gismo as LCD took it over a year ago
- CLHEP for Gismo as used by GLAST today
- CERN’s official version
- LCD Gismo/GismoApps still uses the first version
- Geant requires the official version
stdHEP
- Used almost exclusively in LCD to read and write generator files
- Out of the box, it does not support
- PIC libraries (needed by Root)
- Windows/NT (LCD/SLAC has hacked together its own version)
XML and XMLApps
- We use many independent ASCII input files to Gismo, fastMC (JAS), fastMC (Root), etc. despite the fact that they often contain overlapping information.
- This style of ASCII input tends to be inflexible (to the extent that most of these programs have hard wired some constants rather than change the ASCII file format).
- XML provides a method to build an in-memory database of constants based on one or more flexibly defined ASCII input files.
- XML is becoming a standard.
- XMLApps is an application level library being written by Joanne Bogart to provide access to XML generated database(s).
- Joanne Bogart is predicting a week’s work to complete a single platform Gismo/GismoApps application running with XML/XMLApps (2/3 done already). Unfortunately the week doesn’t start until she returns from vacation.
SIO (Serial Input/Output)
- What XML does for ASCII input files, SIO attempts to do for ASCII output data files
- ASCII output data files tend to be
- Bulky
- Information Lossy
- Inflexible
- SIO addresses these problems by providing
- A binary format (industry standard xdr format)
- A counted, self navigating layout
- Multiple I/O streams each allowing heterogeneous record types
- Enforced data versioning
- Tony Johnson is currently trying to read an SIO file generated from a Gismo/GismoApps application
- Could provide a solution for Geant persistency
Short Term Plan (ɞ months)
- Distribute! (~week(s))
- Take the results of this meeting to create new detectors
- Incorporate all the technical fixes we know about
- Construct the simulation executables (Gismo but not Geant)
- Test
- Distribute code to collaborators
-
- Transmit the output files to Penn.
- Tradeoffs
- Incorporate SIO? (probably). XML? (probably not). Realistic field maps? (probably not).
- How much testing?
Long Term Plan (> 6 months)
- What we are targeting
- Adopt Geant (to replace Gismo)
- Carry forward our GismoApps code to GeantApps
- If XML didn’t make it into the Gismo round, adopt it now
- If SIO didn’t make it into the Gismo round, adopt it now (or come up with another solution to the Geant persistency problem)
- Incorporate realistic magnetic field maps
- Open questions
- Geant persistency model
- {your question here}