LCD simulation software efforts at NIU

1
LCD simulation software efforts at NIU
Dhiman Chakraborty dhiman_at_fnal.gov Northern
Illinois University (NIU)/ Northern Illinois
Center for Accelerator and Detector Development
(NICADD) http//nicadd.niu.edu/ ALCPG
workshop U. Of Texas, Arlington, 09-11 Jan, 2003
2
LC detector simulation software work in progress

• Transition from GISMO to GEANT4,
• Modular, flexible design towards a universal
  package
• Standardize interfaces to offer the best of all
  implementations currently available for
• detector geometry description
• input data (event generator output)
• detector simulator output

3
The LCDG4 detector simulation package(M. Arov,
R. McIntosh, V. Zutshi, D. Chakraborty,
NIU/NICADD)

• A GEANT4-based simulation program
• Not tied to any specific platform (ROOT/JAS/PAW)
• XML description of detector geometry
• Needs structural improvements for better
  generality
• Reads input data in STDHep format
• SIO/ROOT/ASCII output
• S(erial)IO compatible with JAS-based analysis
  code
• LCDG4 can write and JAS can read ROOT files
• Beta release imminent
• 2 known problems fixed during the NIU workshop
• The plan is to merge LCDG4 and MOKKA into a
  single package that combines the best of both.

4
Decoupling the simulator from Root

• LCDG4 is adapted from LCDRoot (M. Iwasaki, T.
  Abe),
• Root internal classes replaced by STL, CLHEP,
• Now a standalone simulation program, not tied to
  any other application/analysis environment,
• I/O compatible with the SLAC/HEP.LCD library
  JAS,
• Root output capability is preserved as an option.

5
(No Transcript)
6
Advantages of the SIO format

• (sequential) access to very large files without
  overburdening memory,
• Allows simultaneous writing to multiple streams,
• Has built-in file compression, which is important
  in view of large event sizes.

7
Non-projective and non-rectangular cell
geometries (in the digitizer)

• Gismo-based simulator only allows towers of
  constant (??,?f),
• An entirely projective tower design is not
  optimal for an E-flow calorimeter,
• (r,?) range is too large for uniform scaling.
• Particular non-projective and non-rectangular
  geometries have been simulated in LCDG4 although
  more work is needed to make it more general and
  flexible,
• Projective rectangular geometries are supported
  just as in Gismo.

8
NIU Prototype
Z(k)
(j,k)
Layer (i)
(0,2)
Cell (j, k)
(-1,1)
(1,1)
f(j)
(0,0)
(-1,-1)
(1,-1)
(0,-2)
The Cell neighborhood
9
Preliminary test results

• Sampling fractions
• Passes order-of-magnitude sanity check, but
  perhaps not quite correct - being checked for
  bugs.

Simulator EM frac Had frac
Gismo 0.02187 0.06338
LCDG4 0.05936 0.07421
10
Total (EMHAD) energy 10 GeV p
entries
1001.0
min
0
max
16.318
mean
9.9900
rms
1.8152
Using LCDG4 with NICADD-designed detector
11
Total HAD energy 20 GeV p
12
Measured energy vs nhits in HCal 20 GeV p
Shape ok, but horizontal scale is 7x Gismo!
13
Hcal longitudinal shower profile of 10 GeV p
14
Total energy measured in Hcal 10 GeV p (1000
events)
15
SIO file contents

• Event header
• Ecal hit list
• Hcal hit list
• Tracker and muon info
• MC particle table

16
How to check it out

• http//nicadd.niu.edu/simulations/software/
• Source binary (Linux) are available
• lcdg4-bin,tar.gz and materials.tar.gz
• the shell command is
• gt lcdg4 b input_file.xml output_file.sio
  Detector_ID MacroFile HepEvt_File

17
Known problems, things to do

• Primaries absent from the MC particle table
• Cause found, figuring out fundamental changes in
  container structures which will optimize speed
  will also cure this problem.
• Too many hits, excess energy may be due to
  overcounting (?), the above procedure should fix
  this as well.
• Geometry is hard-coded in the non-projective
  version. Need to combine with the projective
  version, make it more general and flexible.
• Some more integration/packaging needed.

18
Detector representation issues

• Goals
• portability easy exchange of detector models
• flexibility -- easy changes in detector
  configuration
• uniformity (universality) in detector
  description for
• comparisons and cross checks across detectors.

User
GUI editor
API
Geometry Database (mySQL?)

• Driver API-handler layer

API
API
API
Det. Sim.
Reconstruction
Analysis
19
Summary and plans

• A stand-alone simulation program based on GEANT4
  is ready for testing.
• Order events online at http//nicadd.niu.edu/
• Has all the capabilities of Gismo, plus some
  more.
• A few problems being fixed.
• Emphasis on the need for global integration
  working with SLAC, CALICE, to combine the best of
  Mokka and LCDG4 into one simulation package.
• Figuring out the recipe for detector geometry
  description is a major challenge ahead.
• Start with a lightweight, flexible, functioning
  system.
• Documentation manuals, tutorials, FAQs,
• Its a major undertaking collaborator(s)
  wanted.