Message: Re: Indexing Node Number outside number of slices Not Logged In (login)
 Next-in-Thread Next-in-Thread
 Next-in-Forum Next-in-Forum

More Re: Indexing Node Number outside number of slices 

Forum: Geometry
Re: Question Indexing Node Number outside number of slices (Michael Douglass)
Re: Feedback Re: Indexing Node Number outside number of slices (Gabriele Cosmo)
Date: 08 Dec, 2011
From: Michael Douglass <Michael Douglass>

Hi Gabriele,

Thank you for your suggestion. I have now fixed the geometry and I am in the process of testing it. I will let you know the outcome.

On Thu, 08 Dec 2011 15:12:45 GMT, Gabriele Cosmo wrote:

> Hi Michael,
> 
> you should not use the world volume as mother for your parameterisation
> (this is also specified in the documentation).
> Try introducing an additional box (included in the world with some margin)
> filled with water and have it including the parameterized cubes.
> 
> Cheers, Gabriele
> 
> ------------------
> On Thu, 08 Dec 2011 14:26:31 GMT, Michael Douglass wrote:
> > Hello,
> > 
> > I am experiencing a strange problem in my geant4 code. I have a small
> > cube world volume made of a water material with dimensions 0.72 x 0.72 x
> > 0.72 mm. Inside of this world volume I have 8000 parametrized cubes.
> > Each of these cubes has different dimensions, rotations and randomized
> > positions. The size of these cubes vary in size from 7 - 15 microns. I
> > am confident that none of these geometries are overlapping.
> > 
> > I am experiencing a "fatal error" when running this simulation under
> > windows xp and ubuntu v10. I have run the code through a debugger and
> > have localized the problem to G4ParametrisedNavigation at line 422 of
> > geant4.9.4_p01 in the function ComputeSafety(). The current voxel node
> > number at the point it crashed was 85, but I only have 75 slices in the
> > world volume according to the code.
> > 
> > Some other interesting values that I observed in this code were:
> > pMaxlength = 1.797 e ^(+308)
> > 
> > The pGlobalPoint was (-0.037,-0.115,0.458). The z-coordinate is outside
> > by world volume which has maximum limits of -0.33 mm to 0.33 mm.
> > 
> > The distEndPointSq = (pGlobalPoint-fstependpoint)^2 = 0
> > 
> > Just to be clear, I have not voxellised any of my regions. The
> > voxellisation appears to be an inbuilt geant4 process. I am not using
> > sensitive volumes.
> > 
> > I am using 5 keV gamma particles using the livermore physics list. The
> > track that caused the error appeared to be a secondary electron track
> > which moved beyond the boundries of the world volume.
> > 
> > The problem appears at exactly the same time if I use a fixed starting
> > random seed and at different times when using a random seed. Each
> > simulation ends with the words "Segmentation Fault". I assume this is to
> > the fact that it is trying to access element 85 of the voxel slice
> > matrix which only has 75 slices. (fVoxelHeader).
> > 
> > I read in a geant4 presentation on the Internet that it is possible to
> > delete voxel optimisation and this may fix the problem. However, I am
> > not familiar with this concept.
> > 
> > I am currently attempting to increase the size of the world volume to
> > see if that fixes the problem. I believe this may just cover up the
> > problem if it does fix the problem. I have also tried adjusting the
> > energy cut values which did not fix the problem, although the error
> > occurred at a different track.
> > 
> > I would appreciate any advice that people have to offer.
> > 
> > Thank you
> > 
> > Kind Regards
> > 
> > Michael Douglass
> > 
> 

 Add Message Add Message
to: "Re: Indexing Node Number outside number of slices"

 Subscribe Subscribe

This site runs SLAC HyperNews version 1.11-slac-98, derived from the original HyperNews


[ Geant 4 Home | Geant 4 HyperNews | Search | Request New Forum | Feedback ]