Samplers cause tracking warning

[Laurie Nevay]

In the lhc lattice for generating optics, samplers are attached to everything. Occasionally, warnings are produced:

 -------- WWWW ------- G4Exception-START -------- WWWW -------
 *** G4Exception : GeomField1001
       issued by : G4MagInt_Driver::AccurateAdvance()
       Proposed step is zero; hstep = 0 !
  *** This is just a warning message. ***
  -------- WWWW -------- G4Exception-END --------- WWWW -------

Turning off samplers stops this. I've tried varying aperture types, investigating fields etc. It's samplers. I've tried varying sampler offset from end.

It occurs with a variety of components.

This ultimately slows the simulation down to impractical levels.

[Jochem Snuverink]

A solution proposed by one of the main developers on the hypernews forum is to create our own ExceptionHandler, basically a copy of the Geant4 one, in which one can disable individual warnings. See http://hypernews.slac.stanford.edu/HyperNews/geant4/get/phys-list/913/1.html

This will only suppress the warning, I am not sure if this solution will speed up the simulation again.

[Laurie Nevay]

That's quite interesting - I'll have a read.

I see this at a much lower level now due to changes to beam line padding and sampler placement, but it still exists at the frequency of around 20 times per primary per turn. It's a bit unexpected as it always happens while moving into a sampler but the warning is from the Mag Integrator Driver that only applies to regions with field. I will look at the sampler spacing again.

[Laurie Nevay]

User limits were introduced to prevent steps longer than the boundary of a volume (not default geant4 behaviour) that often resulted in bad tracking and navigation.

Padding was added between all beam line components and samplers.

Curvilinear geometry no longer affects tracking steps.

Geometry tolerance was adjusted to 1nm and length safety was ensured to be the same or larger.

These all combined to remove these warnings. I no longer see these, so I'm closing the issue.