1.1.4. Running in Serial and Parallel

The foundational calculations are deliberately small and should first be run with one MPI rank.

1.1.4.1. Run the Foundations regression examples

From the repository root, run:

test/run_tests -d doc/source/tutorials/foundations --engine=jupyter -j 1 -v 1

To execute an individual notebook interactively, start Jupyter from an environment in which the pyopensn package is installed and open the notebook normally.

1.1.4.2. Launch an input with MPI

1.1.4.2.1. Installed Python package

When the pyopensn package is installed in the active Python environment, launch an input script on four ranks with:

mpiexec -n 4 python input.py

1.1.4.2.2. OpenSn console mode

The same input can be run through the OpenSn console without installing pyopensn into the active Python environment. If the OpenSn executable is on PATH, run:

mpiexec -np 4 opensn -i input.py

The number of ranks changes how the mesh is partitioned, but it should not change the physical problem being solved.