| Summary: | Silence G4Navigator::ComputerStep() when verbose = 0 | ||
|---|---|---|---|
| Product: | Geant4 | Reporter: | Kevin Lynch <krlynch> |
| Component: | geometry/navigation | Assignee: | Gabriele Cosmo <Gabriele.Cosmo> |
| Status: | RESOLVED INVALID | ||
| Severity: | enhancement | ||
| Priority: | P5 | ||
| Version: | 9.2 | ||
| Hardware: | All | ||
| OS: | All | ||
Such warning issued by G4Navigator is intentionally visible since it identifies a real anomaly in the geometry. Coincident surfaces should NOT cause such behavior, unless an overlap due to rounding in the transformations for the positioning of the volumes is generated, and you can verify this by means of the tools available. Stuck tracks cause unphysical pushes by the navigator and eventually abortion of events. I would suggest you to correct this directly in your geometry, eventually correcting possible overlaps by introducing a tiny gap (1 micron or less) between the volumes you identified as problematic. |
Many of the warning cases in G4Navigator::ComputerStep() ignore the value of /geometry/navigator/verbose, and print out large numbers of very large warnings. In my particular case, I'm getting lots of WARNING - G4Navigator::ComputeStep() Track stuck, not moving for 10 steps in volume -VacuumChamberRegion[01]- at point (5589.924214,-22.87434067,4479.188181) direction: (-0.6245245707,-0.6083268471,0.4898035399). Potential geometry or navigation problem ! Trying pushing it of 9e-10 mm ... This particular warning is generated near line 950 of geometry/navigation/src/G4Navigator.cc. This is conditionally compiled by testing G4VERBOSE, but it does NOT check fVerbose at runtime. There are a large number of such cases in this member function. In my case, I know why I'm getting this warning, and it's relatively benign (coincident surfaces from neighboring volumes). I could rebuild my Geant4 distribution, undefing G4VERBOSE, but that's not possible for all users, and has the bad side effect of killing off many other useful warnings/errors elsewhere. It would be nice to truly silence at least these warnings via the Messenger interface, by making them check fVerbose.