Detector Description:
Visualization Attributes Optimisation & Debugging techniques
[Link]
PART IV
Visualization attributes GGE & geometry tree
Each logical volume can have associated a G4VisAttributes object
Visibility, visibility of daughter volumes Color, line style, line width Force flag to wire-frame or solid-style mode
For parameterised volumes, attributes can be dynamically assigned to the logical volume Lifetime of visualization attributes must be at least as long as the objects theyre assigned to
Detector Description: Visualization, optimisation & debugging - Geant4 Course
Each G4VHit concrete class must have an implementation of Draw() method.
Colored marker Colored solid Change the color of detector element
G4Trajectory class has a Draw() method. Blue : positive Green : neutral Red : negative You can implement alternatives by yourself
4
Detector Description: Visualization, optimisation & debugging - Geant4 Course
Implemented in JAVA, GGE is a graphical geometry editor compliant to Geant4. It allows to:
Describe a detector geometry including:
materials, solids, logical volumes, placements
Graphically visualize the detector geometry using a Geant4 supported visualization system, e.g. DAWN Store persistently the detector description Generate the C++ code according to the Geant4 specifications As part of the MOMO Java environment suite
geant4/environments/MOMO/[Link]
5
GGE is provided as a separate tool in Geant4
Detector Description: Visualization, optimisation & debugging - Geant4 Course
Built-in commands defined to display the hierarchical geometry tree
As simple ASCII text structure Graphical through GUI (combined with GAG) As XML exportable format As an additional graphics driver
Implemented in the visualization module
G3 DTREE capabilities provided and more
Detector Description: Visualization, optimisation & debugging - Geant4 Course
Detector Description: Visualization, optimisation & debugging Geant4 Course
PART IV
Smart voxels
For each mother volume
a one-dimensional virtual division is performed
the virtual division is along a chosen axis the axis is chosen by using an heuristic
Subdivisions (slices) containing same volumes are gathered into one Subdivisions containing many volumes are refined
applying a virtual division again using a second Cartesian axis the third axis can be used for a further refinement, in case
Smart voxels are computed at initialisation time
When the detector geometry is closed Do not require large memory or computing resources At tracking time, searching is done in a hierarchy of virtual divisions
Detector Description: Visualization, optimisation & debugging - Geant4 Course 9
Some geometry topologies may require special tuning for ideal and efficient optimisation
for example: a dense nucleus of volumes included in very large mother volume
Granularity of voxelisation can be explicitly set
Methods
Set/GetSmartless()
from
G4LogicalVolume
Critical regions for optimisation can be detected
Helper class G4SmartVoxelStat for monitoring time spent in detector geometry optimisation
Automatically activated if
Percent ------91.70 8.30 Memory -----1k 0k Heads ----1 1 Nodes ----50 3
/run/verbose
Pointers -------50 4
greater than
Total CPU --------0.00 0.00
Volume ----------Calorimeter Layer
Detector Description: Visualization, optimisation & debugging - Geant4 Course
10
The computed voxel structure can be visualized with the final detector geometry Helper class G4DrawVoxels Visualize voxels given a logical volume
G4DrawVoxels::DrawVoxels(const G4LogicalVolume*)
Allows setting of visualization attributes for voxels
G4DrawVoxels::SetVoxelsVisAttributes()
useful for debugging purposes Can also be done through a visualization command at run-time:
/vis/scene/add/logicalVolume <logical-volume-name> [<depth>] Detector Description: Visualization, optimisation & debugging - Geant4 Course 11
Detector regions may be excluded from optimisation (ex. for debug purposes)
Optional argument in constructor of through provided set methods
SetOptimisation/IsToOptimise()
G4LogicalVolume
or
Optimisation is turned on by default
Optimisation for parameterised volumes can be chosen
Along one single Cartesian axis
Specifying the axis in the constructor for
G4PVParameterised
Using 3D voxelisation along the 3 Cartesian axes
Specifying in kUndefined in the constructor for
G4PVParameterised 12
Detector Description: Visualization, optimisation & debugging - Geant4 Course
PART IV
Debugging tools
Optional checks at Construction DAVID Run-time commands OLAP
protruding
overlapping
An overlapping volume is a contained volume which actually protrudes from its mother volume
Volumes are also often positioned in a same volume with the intent of not provoking intersections between themselves. When volumes in a common mother actually intersect themselves are defined as overlapping
Geant4 does not allow for malformed geometries The problem of detecting overlaps between volumes is bounded by the complexity of the solid models description Utilities are provided for detecting wrong positioning
Graphical tools Kernel run-time commands
Detector Description: Visualization, optimisation & debugging - Geant4 Course 14
Constructors of G4PVPlacement and G4PVParameterised have an optional argument pSurfChk:
G4PVPlacement(G4RotationMatrix* pRot, , G4bool pSurfChk=false);
If this flag is true, overlap check is done at construction
A number of points (1000 by default) are randomly sampled on the surface of the volume being created Each of these points are examined if outside of the mother volume, or if inside of already existing other volumes in the same mother volume NOTE: this check may requires lots of CPU time
Depending on the complexity of geometry
Can also be forced on a specific physical volume though the method:
G4bool CheckOverlaps(G4int points=1000, G4double tol=0, G4bool verbose=true);
Worth to try when first implementing a geometry of some complexity !
Detector Description: Visualization, optimisation & debugging - Geant4 Course 15
DAVID is a graphical debugging tool for detecting potential intersections of volumes Accuracy of the graphical representation can be tuned to the exact geometrical description.
physical-volume surfaces are automatically decomposed into 3D polygons intersections of the generated polygons are parsed. If a polygon intersects with another one, the physical volumes associated to these polygons are highlighted in color (red is the default).
DAVID can be downloaded from the Web as external tool for Geant4
[Link]
Detector Description: Visualization, optimisation & debugging - Geant4 Course
16
Built-in run-time commands to activate verification tests for the user geometry. Tests can be applied recursively to all depth levels (may require CPU time!): [recursion_flag] geometry/test/run [recursion_flag] or
geometry/test/grid_test [recursion_flag] to start verification of geometry for overlapping regions based on a standard grid setup geometry/test/cylinder_test [recursion_flag] shoots lines according to a cylindrical pattern geometry/test/line_test [recursion_flag] to shoot a line along a specified direction and position geometry/test/position and geometry/test/direction to specify position & direction for the line_test
Resolution/dimensions of grid/cylinders can be tuned
Detector Description: Visualization, optimisation & debugging - Geant4 Course 17
Example layout:
GeomTest: no daughter volume extending outside mother detected. GeomTest Error: Overlapping daughter volumes The volumes Tracker[0] and Overlap[0], both daughters of volume World[0], appear to overlap at the following points in global coordinates: length (cm) ----- start position (cm) ----- ----- end position 240 -240 -145.5 -145.5 0 -145.5 Which in the mother coordinate system are: length (cm) ----- start position (cm) ----- ----- end position . . . Which in the coordinate system of Tracker[0] are: length (cm) ----- start position (cm) ----- ----- end position . . . Which in the coordinate system of Overlap[0] are: length (cm) ----- start position (cm) ----- ----- end position . . .
(list truncated) (cm) -----145.5 (cm) -----
(cm) -----
(cm) -----
Detector Description: Visualization, optimisation & debugging - Geant4 Course
18
Adopt tracking of neutral particles to verify
boundary crossing in opposite directions Stand-alone batch application
Provided as extended example Can be combined with a graphical environment and GUI ex. Qt library Integrated in the CMS Iguana Framework
Detector Description: Visualization, optimisation & debugging - Geant4 Course
19
Detector Description: Visualization, optimisation & debugging - Geant4 Course
20
