Description

We have an increasing number of use cases where want to use functionality from the OpenMC shared library without actually running OpenMC itself. This usually involves creating a temporary directory, exporting a Model to XML, initializing the OpenMC shared library via openmc.lib.init, and then calling the function(s) of interest from openmc.lib. Examples of this include:

• MicroXS.from_multigroup_flux
• WeightWindowsList.export_to_hdf5
• Mesh.material_volumes

This PR introduces a new openmc.lib.TemporarySession context manager class that makes this process a little easier by combining the first few steps together (moving to a temp dir, exporting XML, calling openmc.lib.init). So the first line of code under the with statement can be a call to something from openmc.lib. I've updated each of the methods above to use the new TemporarySession class.

Use for MicroXS

For some depletion/activation workflows, the user may want to make repeated calls to MicroXS.from_multigroup_flux. Right now, this will cause openmc.lib.init to get called (hence loading cross sections) each time, which may result in long execution times. I've added a new session argument on from_multigroup_flux that allows one to use the same temporary session so that openmc.lib only needs to be initialized once. So it might look like:

with openmc.lib.TemporarySession() as session:
    for mat in long_list_of_materials:
        micros = MicroXS.from_multigroup_flux(energy, flux, nuclides=mat.get_nuclides(), session=session)

Chain caching

In the above example, another issue with repeated calls to MicroXS.from_multigroup_flux is that a chain file will get loaded from XML each time. Thanks to PR #3436 by @shimwell, this could be avoided by creating a Chain object and then passing it to the chain_file argument. In this PR, I've also made things a little bit smoother by caching Chain instances behind the scenes so that a user can just set openmc.config['chain_file'] instead of having to pass around the Chain object.

Checklist

• I have performed a self-review of my own code
• I have run clang-format (version 15) on any C++ source files (if applicable)
• I have followed the style guidelines for Python source files (if applicable)
• I have made corresponding changes to the documentation (if applicable)
• I have added tests that prove my fix is effective or that my feature works (if applicable)