Efficient Virtual Segmentation for Large-scale Detector Systems
undefined
Sensitive Detector Segmentation
Norman Graf (SLAC)
LC-ECFA Meeting, DESY
May 28, 2013
2
Problem Statement
LC Detectors being considered for the ILC and CLIC are
highly segmented, resulting in hundreds of millions to
billions of readout channels.
Implementing each readout channel as a separate
volume in Geant4 is impractical.
Define Sensitive Detectors as larger volumes and use
“virtual segmentation” to define readouts, e.g.
Tracking sensitive detector is silicon wafer, virtual segmentation returns pixels
or strips.
TPC sensitive detector is full endplate, virtual segmentation returns pads
Calorimeter sensitive detector is scintillator sheet, or gas volume, virtual
segmentation returns cells.
Functionality
Given local position, return cell ID
Given cell ID, return local position
Given cell ID, return list of neighboring cell IDs
Return cell size (allows energy density clustering)
3
Dependencies
Need to have a separate package containing
segmentation classes on which both the
simulation and reconstruction packages depend.
Do not want to couple the reconstruction to
Geant4.
Also don't want to have two separate
implementations of segmentors.
4
slic
G4SD
binding
Geant4
Segmentation
Geometry
reco
?
lcdd
Cartesian Grid
For each grid
x
deltaX
y
deltaY
Allows for effective
gaps in between pads
R-Phi Readout, e.g. BeamCal
7
For each annulus:
rmin
rmax
nCells
phi0
deltaPhi
Allows for effective
gaps in
both radius and phi
Allows
for staggered cells
Hexagons
Define hexagon “radius”
Can define effective gaps between cells using
two “radii”
8
Attaching Segmentation Class to Volume
Need the ability to specify location of origin and
orientation of segmentation coordinates with
respect to sensitive geometric volume.
9
Issues
In calorimeters, how to handle Geant4 steps
which straddle cell borders or cross “gaps”
How to handle edges of segmentation classes
irregular cell sizes and shapes
missing neighbors
Semi-Digitization
How to handle position-sensitive digitization such as
charge-sharing across boundaries (e.g. RPC) or
efficiency of light collection in scintillator readout.
10
Software Architecture
Segmentation classes should be closely related
to Geant4 primitive volumes for efficiency in
defining, implementing and utilizing them.
Do not want reconstruction to depend on Geant4
Standalone package upon which both simulation
and reconstruction classes depend.
Runtime binding via plugin mechanism might
lead to problems of provenance.
Prefer compile-time binding of classes, run-time
definitions for parameters.
11
Highly segmented detectors with millions to billions of readout channels require a practical approach for simulation and reconstruction. Implementing each channel separately is impractical, necessitating the use of virtual segmentation to define readouts for different detector components. Dependencies, functionality requirements, and specific segmentation strategies like Cartesian grids, R-Phi readouts, and hexagons are addressed in this content.
Download Presentation
Please find below an Image/Link to download the presentation.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author. Download presentation by click this link. If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.
E N D
Presentation Transcript
Sensitive Detector Segmentation Norman Graf (SLAC) LC-ECFA Meeting, DESY May 28, 2013
Problem Statement LC Detectors being considered for the ILC and CLIC are highly segmented, resulting in hundreds of millions to billions of readout channels. Implementing each readout channel as a separate volume in Geant4 is impractical. Define Sensitive Detectors as larger volumes and use virtual segmentation to define readouts, e.g. Tracking sensitive detector is silicon wafer, virtual segmentation returns pixels or strips. TPC sensitive detector is full endplate, virtual segmentation returns pads Calorimeter sensitive detector is scintillator sheet, or gas volume, virtual segmentation returns cells. 2
Functionality Given local position, return cell ID Given cell ID, return local position Given cell ID, return list of neighboring cell IDs Return cell size (allows energy density clustering) 3
Dependencies Need to have a separate package containing segmentation classes on which both the simulation and reconstruction packages depend. Do not want to couple the reconstruction to Geant4. Also don't want to have two separate implementations of segmentors. 4
slic reco G4SD binding lcdd Geometry ? Geant4 Segmentation
Cartesian Grid For each grid x deltaX y deltaY Allows for effective gaps in between pads
R-Phi Readout, e.g. BeamCal For each annulus: rmin rmax nCells phi0 deltaPhi Allows for effective gaps in both radius and phi Allows for staggered cells 7
Hexagons Define hexagon radius Can define effective gaps between cells using two radii 8
Attaching Segmentation Class to Volume Need the ability to specify location of origin and orientation of segmentation coordinates with respect to sensitive geometric volume. 9
Issues In calorimeters, how to handle Geant4 steps which straddle cell borders or cross gaps How to handle edges of segmentation classes irregular cell sizes and shapes missing neighbors Semi-Digitization How to handle position-sensitive digitization such as charge-sharing across boundaries (e.g. RPC) or efficiency of light collection in scintillator readout. 10
Software Architecture Segmentation classes should be closely related to Geant4 primitive volumes for efficiency in defining, implementing and utilizing them. Do not want reconstruction to depend on Geant4 Standalone package upon which both simulation and reconstruction classes depend. Runtime binding via plugin mechanism might lead to problems of provenance. Prefer compile-time binding of classes, run-time definitions for parameters. 11
