in an attempt to simulate pmts in a water cerenkov tank, G4Sphere was used to
define the solid volume and the corresponding logical volumes were registered as
sensitive volumes. however, no hits were registered (in ProcessHits()) even when
a particle was shot right at the volume. visualization was also used to check to
see if it was hit.
the solid was changed to G4Box and G4Tubs, with the logical and physical volume
code unchanged. in both cases hits were registered, even when a geantino was
used as a test particle.
in a UserSteppingAction, the volumes were checked from the fTrack info. the
sensitive volume only showed up in the latter two cases. for example, in the
case of the geantino as primary the output was essentially :
for Box, Tubs ->  World, pmt1, World
for Sphere    ->  World
a "stripped down" version of the code is at
http://hepnt.physics.neu.edu/staff/mccauley/geant.html
the solids can be changed in TankDetectorConstruction.cc
the sim runs in batch and calls the mac file tank.mac