Hi there,

We've found in the wild, while running G4 9.6 patch2, instances of the do/while loop in G4MuMinusCapturePrecompound::ApplyYourself causing infinite loops.  The code I'm referring to is here:

http://www-geant4.kek.jp/lxr/source/processes/hadronic/stopping/src/G4MuMinusCapturePrecompound.cc#L91

The stack we have from GDB is:

#0  0x00002b8bb8e25bb7 in CLHEP::dSFMTEngine::flat() ()
   from /cvmfs/atlas.cern.ch/repo/sw/software/x86_64-slc5-gcc43-opt/17.7.1/AtlasProduction/17.7.1.2/InstallArea/x86_64-slc5-gcc43-opt/lib/libAtlasCLHEP_RandomGenerators.so
#1  0x00002b8bb3603b9e in G4RandomDirection () at /tmp/jchapman/build/source/global/HEPRandom/include/G4RandomDirection.hh:57
#2  G4MuMinusCapturePrecompound::ApplyYourself (this=0x20419800, projectile=..., targetNucleus=...) at src/G4MuMinusCapturePrecompound.cc:185
#3  0x00002b8bb35fa82c in G4HadronStoppingProcess::AtRestDoIt (this=0x2065d000, track=...) at src/G4HadronStoppingProcess.cc:194

Coming back later - with the simulation having not completed a step yet - gives:

#0  operator* (a=<optimized out>, p=...) at /tmp/jchapman/CLHEP/Vector/ThreeVector.icc:55
#1  CLHEP::HepLorentzVector::boostVector (this=0x7fff5d6d0db0) at /afs/cern.ch/sw/lcg/external/clhep/2.1.2.3-atl01/src/CLHEP/Vector/src/LorentzVector.cc:188
#2  0x00002b8bb3603adb in G4MuMinusCapturePrecompound::ApplyYourself (this=0x20419800, projectile=..., targetNucleus=...) at src/G4MuMinusCapturePrecompound.cc:177
#3  0x00002b8bb35fa82c in G4HadronStoppingProcess::AtRestDoIt (this=0x2065d000, track=...) at src/G4HadronStoppingProcess.cc:194

Looks and smells like a bug to me.  When the code fails, it is propagating a mu- that begins with 116.5402 MeV of KE (pT=118.7672) inside our straw tube tracker (mostly light materials).  Given that it's trying to stop, I suppose the KE has run out of the muon already.


Thanks,
Zach