In the basic CO2 Injection example provided in the documentation, it is suggested that the simulation can be launched using 4 MPI processes with the following command:

mpirun -np 4 geosx -i simpleCo2InjTutorial_smoke.xml -x 1 -y 1 -z 4

However, the internal mesh defined in the input file is as follows:

<InternalMesh
  name="cartesianMesh"
  elementTypes="{ C3D8 }"
  xCoords="{ 0, 1000 }"
  yCoords="{ 450, 550 }"
  zCoords="{ 6500, 7700 }"
  nx="{ 50 }"
  ny="{ 1 }"
  nz="{ 150 }"
  cellBlockNames="{ cellBlock }" />

Here, nz = 150 is not divisible by 4, which leads to uneven domain partitioning when using -z 4. This results in a very slow simulation or even stalls with no clear feedback to the user, which can be confusing for beginners.

Please revise the documentation to either recommend a process count that evenly divides the domain (e.g., -z 3 or -z 5 for this mesh), or explain the importance of matching the domain decomposition (-x -y -z) to the mesh resolution (nx ny nz), and mention potential issues that may arise from unbalanced partitioning.

Thank you for your great work on GEOSX!