I have found a problem when positioning multiple G4IntersectionSolid volumes within a mother volume. The problem is there both in G4 v.8.0 and v.7.1. In my setup I want to position several fibers within a slice of scintillator. The scintillator slice is a cylindrical sector and the fibers are evenly placed along radiuses of the cylinder. To create fibers with the correct shape I use thin cylinders and cut their end faces with a rounded surface identical to the surface of the main slice. This is done by intersecting the fiber cylinder with a cylindrical sector of the same shape as the scintillator slice. You can find some example code which should work fine with the out-of-the-box novice example 02 in /afs/cern.ch/user/l/leonardi/public/geant4/ExN02DetectorConstruction.hh /afs/cern.ch/user/l/leonardi/public/geant4/ExN02DetectorConstruction.cc The problem I have is that if I position up to 3 fibers (using the Veto_NWLSFiberPhi parameter in the code) everything runs smoothly, but if I use 4 or more fibers the whole program crashes with a core dump in the geometry optimization sector when I issue the /run/beamOn command. The crash point obtained by gdb can be found in /afs/cern.ch/user/l/leonardi/public/geant4/CrashPosition.txt Visual inspection of the slice with any number of fibers does not show any anomaly when using more than 3 fibers. Is there a problem in the way I use intersection solids or is this a problem with G4? Thanks Emanuele Leonardi
The fix was directed to ignore the optimisation for empty intersections.