Unexpected NaN Values in Particle Position Computation

[Itay-max]

Bug Description

Unexpected behavior in particle position computation leading to NaN values.

Steps to Reproduce

1. Run a simple simulation using the latest OpenMC version.

Simulation example

• Geometry Description: Water-filled sphere centerd at the origin surrounded by a vacuum region.
• Materials Used: The only material defined is water.
• Particle Source: An isotropic point source of neutrons is placed at the origin (0, 0, 0) with an energy distribution defined as a discrete energy of 1e6 MeV.
• The simulation is set to run in fixed source mode with 1,000,000 particles and 100 batches.

2. Include the following modification (under this line in particle.cpp file):

   if ((this)->E_last()*1e8 < 10)
   {
       std::cout << "p->E_last()  "<< (this)->E_last() <<std::endl;
       std::cout << "p->r() "<< this->r().x<<" "<<this->r().y<<" "<<this->r().z<<" "<<std::endl;
   }

3. Recompile the project:

cmake ../openmc          
   -DCMAKE_CXX_COMPILER=mpicxx    
   -DOPENMC_USE_MPI=on         
   -DHDF5_PREFER_PARALLEL=on 
   -DCMAKE_INSTALL_PREFIX=/root/OpenMC/install 

make install

Observed Behavior

The code prints nan values for the position of the particle the once in around 15 batches. Here is an example for such case:

p->E_last()  4.33725e-08
p->r() nan nan nan

Expected Behavior

The particle position should be well-defined and should not result in NaN values.

Environment

OpenMC official Docker image.

[Itay-max]

Might be related: https://openmc.discourse.group/t/segmentation-fault-on-simple-model-after-many-particles/3250

[pshriwise]

Just noting this comment on the geometry description this morning:

Geometry Description: Water-filled sphere centerd at the origin surrounded by a vacuum region.

@Itay-max, is this based on the same geometry as the one @shimwell provided in the forum discussion you've linked above?