Upcoming release of OpenMC 0.6.1

[Paul Romano]

OpenMC developers,

I created a release-0.6.1 branch for final testing before the actual release, which I'd like to get out in the next week. If you have a few moments, please check out the new branch and confirm that you have no issues with it.

Most of you are probably unaware that OpenMC releases automatically have a DOI assigned to them now, e.g. see http://dx.doi.org/10.5281/zenodo.10719. I've decided to forgo the GitHub pre-release feature as it prevents Zenodo from issuing a DOI for the final release.

Best regards,

Paul

[Jon Walsh]

I haven't looked into it yet, but I'm getting a seg fault after ~1E9 particles with a model of core IIIb from the B&W 1484 series.  I've run a bunch of other very similar models without issue with this same pre-release version of the code.  Also, this same model has run to completion with previous releases of the code.

Jon

[Paul Romano]

You can try compiling with cmake -Ddebug=on and then doing a restart with the particle restart file (openmc -r particle_530_808676.h5). If that works, at least we'll know where the code is segfaulting.

[Jon Walsh]

Below are two errors/warnings: the first is the error that seg faulted the code when running the whole simulation and the second is the warning that came from just the restart - but did not kill the simulation.  Interesting directional cosine, though.  Let me know if this reeks of bad geometry input.  As far as I can tell, I have vacuum BC's on every outer surface.

So here is the runtime error when I ran the entire simulation:

Program received signal SIGFPE: Floating-point exception - erroneous arithmetic operation.                                              
                                                                                                                                        
Backtrace for this error:                                                                                                               
#0  0x2B01F2940457                                                                                                                      
#1  0x2B01F2940A6E                                                                                                                      
#2  0x3050C302CF                                                                                                                        
#3  0x5822BC in __mesh_MOD_get_mesh_indices                                                                                             
#4  0x54348F in __tally_MOD_score_tl_on_mesh                                                                                            
#5  0x54CE4D in __tally_MOD_score_tracklength_tally                                                                                     
#6  0x46FDD9 in __tracking_MOD_transport                                                                                                
#7  0x4E3854 in __eigenvalue_MOD_run_eigenvalue                                                                                         
#8  0x4EB6FF in MAIN__ at main.F90:0

And here's the error just running the restart:

     Entering cell 45
     Crossing surface 500
 WARNING: After particle 808676 crossed surface 500 it could not be located in
          any cell and it did not leak.
 Neutron 808676              
   Level 0
     Universe         = 0
     xyz =   6.3764E+02  3.9297E+02 -8.8024E+02
     uvw =  -6.7336E-01 -3.6668E-01  1.0174E+00
   Weight = 1.00000       
   Energy = 4.58509E-07

[Paul Romano]

The directional cosine is suspect indeed. I'm wondering if this could be related to issues #91 and #258. A few things to try

• Find all places where uvw is changed and see where the magnitude becomes greater than one.
• Run with geometry debugging mode (-g or --geometry-debug)